TERESA MARYANSKA

ANKYLOSAURIDAE (DINOSAURIA) FROM MONGOLIA (plates 19-36)

Abstract. - The paper deals with Asian Ank ylosauria, assigned to the family Ankylosauridae, BROWN. New material collected by the Polish-Mongolian exped itions to Mongolia in three Late Cretaceous formations (Djadokhta, Barun Goyot and Nernegt) is described. Two new genera and species: Saichania chulsanensis and Tarchia kielanae, both from , the ?Campanian Barun Goyot Formation, are described. Revised diagnoses of all Mongolian ankylosaurid species are given. An ossified conchal system in nasal cavities of Saichania chulsanensis, a unique feature among the extinct lower tetrapods is described. The postfrontal, epipterygoid, and additional ossifications in the antorbital wall closing the orbit in the front, as well as four pairs of secondary ossifications of dermal origine, which cover partly the skull bones, are found in the skull of Pinacosaurus grangeri and Saichania chulsanensis. Presented reconstruction of the hyoid apparatus of Saichania chulsanensis differs from that known in Psittacosaurus and Protoceratops and shows similarities to that of Squamata. In the pectoral girdle of Saichania chulsanensis the fusion of the coracoid with the transverse process of the first dorsal vertebra is stated. A morphological (and possibly also phylogenetical) sequence of Mongolian Ankylosauridae, beginning with an unnamed Early Cretaceous specimen from Khovboor, through Talarurus, Pinacosaurus, Saichania to " Dyoplosaurus" is presented. The position of Ankylosauria within the Ornithischia is discussed. It is presumed that the quadrupedality in Ankylosauria is primary. The origin of Ankylosauria from Pseudosuchia, independent from the line of bipedal and cursorial Ornithopoda seems possible. Common origin of Ankylosauria and Stegosauria is suggested.

Contents

Page Introduction . . 86 Material . .. 93 Measurements. 95 General remarks on the systematics 97 Check list of Asian Ankylosauria 98 Descriptions . 99 Suborder Ankylosauria . 99 Family Ankylosauridae, BROWN 99 Genus Talarurus MALEEV . . . 99 Talarurus plicatospineus MALEEV . 100 Talarurus disparoserratus (MALEEV) 100 Genus Pinacosaurus GILMORE 101 Pinacosaurus grangeri GILMORE 101 Genus Saichania nov. . . . 103 Saichania chulsanensis sp.n. 103 Genus Tarchia nov. . . 105 Tarchia kielanae sp.n, 108 Anatomy . 108 Osteology of the skull 109 Individual bones . . . 109 Accessory elements. . 118 Bony nasal structures 119 Remarks on the nares and nasal cavities 122 86 TER ESA MARYANSKA

Osteology of the postcranial skeleton. 124 Vertebral column and ribs . . 124 Pectoral girdle and fore limb . 127 Pelvic girdle and hind limb .. 130 Dermal armour ...... 131 Geographic and stIatigraphic distribution ot Ankylosauria in Asia 132 Evolution of Asiatic Ankylosauridae . . . . . 134 Position of Ankylosauria within Ornithischia . 136 Habits of Ankylosauridae 139 References ...... 140

INTRODUCTION

Ankylosauria from the Cretaceous of Asia have been studied only preliminarily. A mono­ graph of COOMBS (MS) devoted to ankylosaurs contains few data about Asiatic representatives. The first note on the occurrence of the armored dinosaurs in Asia come from MATLEY (1923) who has described Lametasaurus indicus from the Cretaceous of Jubbulpore (Lameta Beds). Assignment of Lametasaurus to the Ankylosauria has been questioned (CHAKRAVARTI, 1935). Notes by WIMAN (1929) and GILMORE (1933a) pertain to the occurrence of remains of the true Ankylosauria in China. These fragmentary findings did not allow more precise identification. Pinacosaurus grangeri GILMORE described on the basis of a deformed skull (AMNH 6523) from the Djadokhta Formation of Shabarakh Usu (Bayn Dzak) was the first Asiatic species undoubtedly belonging to Ankylosauria (GILMORE, 1933b). In 1935 YOUNG described Pina­ cosaurus ninghsiensis from the Cretaceous of.Ninghsia (China). In the USSR RIABININ (1939) has found ankylosaufs in the Upper Cretaceous near Tashkent. It seems also probable that the remains descr.bed by him as possible Ceratopsia (I.e., table 11, figs I, la, 4, 4a; table 12, figs 4,5; table 14, figs 2, 2a) belong also to the Ankylosauria. On the basis of incomplete and badly preserved materials found during 1930-1931 by the Sino­ Swedish expeditions, BOHLIN (1953) reported the occurrence of the ankylosaurs in several sites of Cretaceous age in the eastern Kansu and has described three new genera and species: Sauroplites scutiger, Heishansaurus pachycephalus and Peishansaurus philemys, as well as a new genus and species Stegosaurides excavatus assigned to ?Stegosauria. MALEEV 1 (1956), ROMER (1966), STEEL (1969) and others assigned Stegosaurides excavatus to the Ankylosauria. Not all species established by BOHLIN can be regarded as valid because of their incomplete type specimens. Data pertaining to occurrence of Ankylosauria in China are supplemented by the reports of the 1960 Sino-Soviet Palaeontological Expeditions (ROZHDESTVENSKY, 1961a, 1961b). Two skeletons of ankylosaurs have been excavated near Maortu (Alashan). This material so far undescribed is housed in Peking. Occurrence of ankylosaurs has also been noted in Inner Mongolia (Young, 1964) and in Nanhsiung (Kwantung) in the Nanhsiung Formation of Late Cretaceous age (YOUNG, 1965; CHENG et al., 1973). Abundant and well preserved remains of ankylosaurs have been excavated from the Upper Cretaceous of Mongolia by Soviet expeditons during 1946-1949 (MALEEV, 1952a, 1952b, 1952c, 1954b, 1956). Remains from Bayn Shireh (Bayn Shireh svita, see p. 132) described by MALEEV (1952b) as Talarurus plicatospineus include numerous specimens housed in the Pa­ laeontological Institute in Moscow. MALEEV (1952a) established a new family, the Syrmosauridae (regarded herein as a synonym of the Ankylosauridae) with one genus Syrmosaurus (type species Syrmosaurus viminicaudus) from the Djadokhta Formation. MALEEV (1952b) has also described Syrmosaurus disparoserratus from Sheeregeen Gashoon (Bayn Shireh svita) and Syrmosaurus sp. from Oshi.

• Referred to in previous publications as MALEYEV. ANK.YLOSAURIDAE FROM MONGOLIA 87

The new ankylosaurs were discovered in the Upper Cretaceous of Mongolia by the Polish­ Mongolian Palaeontological Expeditions of 1963-1971 (KIELAN-JAWOROWSKA & DOVCHIN, 1969; KIELAN-JAWOROWSKA& BARSBOLD, 1972; MARYANSKA 1970, 1971). A rich and interesting collection of ankylosaurs was discovered by the Soviet-Mongolian Expeditions to Mongolia in 1970-1971. The latter collection contains Upper Cretaceous forms form e.g. Alag Teg (TVERDOCHLEBOV & TSYBIN, 1974) and Khermeen Tsav (KRAMARENKO, 1974) as well as Lower Cretaceous forms from the vicinity of Khovboor (KRAMARENKO, l.c.) and (?) Khuren Dukh (KALANDADZE & KURZANOV, 1974). Fragmentary undescribed remains of ankylosaurs from several sites in Kazakhstan (e.g. Shakh-shakh) housed in the Laboratory of Palaeobiology of Kazakh Academy of Sciences in Alma Ata and remains from Yakutia (USSR) housed in the Palaeontological Museum in Moscow complete the collection of ankylosaurs from Asia (see table 1 and text-fig. 1). The present paper is devoted to ankylosaurs collected in the Mongolian People's Republic by Polish-Mongolian Palaeontological Expeditions in 1963-1971, and housed in the Institute of Paleobiology of the Polish Academy of Sciences in Warsaw and in Geological Institute Section

- i :90~ ---- \700 '- 60' /'0' I ! I

027

22

Fig. 1 Diagramatic map of Asia showing the localities in which ankylosaurs were found. 1 - Bayn Dzak, 2 - Alag Teg, 3 ­ AItan Via, 4 - Nemegt, 5 - Khulsan, 6 - Khermeen Tsav, 7 - Sheeregeen Gashoon, 8 - Khovboor, 9 - Bayn Shireh, 10 - Baga Tarjach, 11 - Baysheen Tsav, 12 - NW Ninghsia, 13 - Chia-yii-kuan, 14 - Hui-hui-pu, 15­ Ehr-hia wu-kuan, 16 - Maortu, 17 - Tsondolein Khuduk, 18 - Tebch, 19 - Iren Dabasu, 20 - Khuren Dukh, 21 - Oshi Nur, 22 - Jubbulpore, 23 - Alym Tau, 24 - Shakh-shakh, 25 - Djara Khuduk, 26 - Laiyang, 27- Nanhsiung, 28 - Kempenday. 8R TERESA MARYANSKA

Ta Ankylosaurian

Ide n t i f i c a t i o n Locality (forexpl an ati on Mu seum cat. no of numbers see origin al I in this paper Text-fig. 1)

" Dyoplosaurus" gigantells PIN, undescribed speci- Khermen Tsav (6) MALEEY men , no catalogue number

Dy oplosaurus cf. giganteus (MA- ZPAL MgD-I/43, housed RYANSKA, 1970) "Dyoplosaurus" giganteus no w in GISPS, VIan Bator Allan Via (3) M ALE EY Dyoplosaurus sp. (MARYANS KA, ZPAL MgD-I/42, a nd 1970) MgD-I/49, MgD-I/11 3

Dyoplosaurus giganteus MALE EY " Dyoplosaurus" giganteus PIN 551-29 Nemegt (4) (MALEEY, 1956) MALE EY .

Pinacosauru s (KRAMARENKO, PIN, undescribed speci- 1974) men , no ca ta log ue number Saichania chulsan ensis gen. Khermeen Tsav II (6) n., sp. n. ankylosaurid (KI ELAN-JAwOROW- ZPAL MgD-I/114 SKA & BARSBOLD, 1972)

ankylosaurid (KIELAN-JAwOROW- Sa ichania chulsanensis gen . GI SPS 100 /151 hol ot ype SKA & BARSBOLD, 1972) n., sp. n. Khulsan (5)

ankylosaurid (KI ELAN'-JAWOROW- Tar chia kielanae gen. n., ZPAL MgD-I/111 hol otype SKA & BARSBOLD, 1972) sp. n. - Pinacosaurus ninghsiensis In st. Vert. Paleont., Pek ing , Pinacosaurus grangeri NW of Ninghsia (12) YOUNG (YOUNG, 1935) GILMORE no cat alogue number

PIN, undescribed speci- ankylosaurs KUROCHKIN & K A- Pinaco saurus grangeri men s, no ca ta log ue num- Alag Teg (2) LANDADZE 1970) GILMORE bers

Pinacosaurus grangeri GILMORE AMNH 6523 holot ype (GILMORE , 1933)

Syrmosaurus viminicaudus M A- PIN 614 LEEY (M ALEEY, 1952) Bayn D zak (1) Pinacosaurus grangeri GI L- ZPAL M gD -I1/7 MORE ZP AL MgD-I1 /9 Pinacosaurus grangeri GI LMORE ZPAL MgD-II/31 (MARYANSKA, 1971) ZP AL MgD-II/32 ZPAL MgD-II/27

undescr ibed re ma ins ankylosaurs (ROZHDESTYENSKY, a nkylosaurids hou sed in Paleob iol. Lab. Shakh -shakh (24) 1972) in Alma At a

Nodosauridae gen. et sp . indet. ankylosaurids PIN, no catalogue number Alym Tau (23) (RIABININ , 1939) ANKYLOSAURIDAEFROM MONGOLIA 89 ble 1 remains from Asia

I Formation and/or other Presumable age I Associated vertebrate faun a stratigraphic dat a I Upper white beds of Khe r- I Therizinosaurus cheloniformis Contemporaneous with Nemegt meen Tsav (GRAD zINsKr& Form ati on Tarbosaurus bataar, hadrosaurs, sauropods JER ZY KI EWICZ, 1972)

Nemegt Formation (GRA- Upper Camp anian or Lower Gorgosaurus lancinator, Tarbosaurus bataar, Gal- DZINSKI, 1970; MARTI NSON Maastrichtian (MARYANS KA & limimus bullatus, Deinocheirus miriflcus, Saurolophus et a/., 1969 ; GRADzrNsKI& OSM6LSKA , 1975) Maastrichtian angustirostris, turtles JERZYKI EWrCz, 1972) (MARTINSON, 1975) - Tarbosaurus bataar, Gallimimus bul/atus, Gorgo- Nemegt Formation (GRA- Upper Campanian or Lower saurus novojilovi, Therizinosaurus chelonifo rmis , Ne- DZINSKI, 1970; MARTI N- Maastrichtian (MARYANSKA & megtosaurus mongoliensts, Saurolophus angustirostr is, SON et al., 1969; GRAD ZIN- OSM6 LSK A, 1975) Maastrichtian Prenocephale prenes, Homalocephale calathocercos, SKI & JERZYKIEWrCz, 1972) (MARTINSON, 1975) Paraligat or ancestralis

Baga ceratops rozhdest venskyi, small therapod s, Red beds of Khermeen oviraptors, Cherm insaurus kozlowsk ii, Erdenetesaurus Tsav (GRADZ INSKI & JE- robinsoni, Darchansaurus estesi, M acrocephalosaurus Contemporaneous with Barun RZYKIEWICZ, 1972) Kher- gilmorei and other lizard s. Djado chtath erium catapsa- Goyot Formation (KIEL AN-JAwO- meen Tsav formation Joides, Ne megtbaatar gob iensis, Barunlestes butleri, ROWSKA, 1975a) (KIELAN-JAWOROWSKA . Asiory ctes nemegtensis, Deltatheridium pretrituber- 1975a) cu/are tardum and othe r mammals

?Proto ceratops ko zlowskii, Tylocephale gilmorei, Ve- lociraptor sp., carnosaur indet ., sauropod indet., Barun Go yot Formation Uppe r Sant onian -Camp anian Zangerlia testudinimorpha, Gobipterix minuta, Na- (MARTINSON et a/.• 1969 ; (M AR1I NSON, 1973) ?Middle Cam- ransaurus chulsanensis, Macrocephalosaurus and GRADZINSKI &JERZYKIE- panian (KrELAN-JAWORowSKA, other lizards , Djadochtatherium catapsa/oides, Ne- WICZ. 1974) 1974b, 1975a , b) megtbaatar gobiensis, Barun/estes butleri, Asioryc tes nemegetensis and other mammals

- - ?Contemporaneous with Dja- dokhta Formation

unnamed formation (TVER- DOCHLEBOV and TSYBIN, ?Contemporaneous with Dja- carn osaurs, sauropods turtl es 1974) Barun Goyot svita dokhta Fo rmation (SHUVALOV, 1975)

Protoc eratops andrewsi, Ve/ociraptor mongoliensis, S auronithoides mongoliensis, Oviraptor philoceratops, Upper Santonian-Ca mpanian (RO-I Hadrosauridae indet., Shamosuchus aja dochtaensis, Diadokhta Formation (BERK EY&MORRIS, 1927) zhdestvensky, 1971) Gobiosuchus kie/anae, Adamisaurus magn ident atus, ?Santonian (KIELA N-JAWOROW- LEFELD, 1972) Barun Goyot Macrocephalosaurus fe rrugenous and other lizard s, SKA 1974b) Turonian-Lower Se- svita (SHUVALOV. 1975; Djadochtatherium matth ewi, Kryptobaatar dashevegi, nonian (SOCHAVA, 1975) Cam- SOCHAVA, 1975) Deltatheridium pretr ituberculare pretritubercu/are, pan ian -Maastrichtian (SHUVALOV Za/ambda/estes lechei and other mammals 1975)

Aralosaurus tuberifrons, carnosaurs, sauropods. Beleutin ska svita (ROZH- Turonian (ROZHDESTVENSKY , turtles, Kansajsuchus borealis DESTVENSKY, 1964) 1974)

Jaxartosaurus aralensis, thero pods, sauropods Beleutinska svita (ROZH- Coniacian (ROZHDESTVENS KY, DESTVENSKY, 1971) 1974) 90 TERESA MARYANSKA - _._- - Ident if i cation Localit y (for explanation Museum cat. no of numbers see Text- original in this paper fig. I) _. I i Ankylosauria (ROZHDESTV ENSK Y, ankylosaurs PIN, no catalogue number Djara-Khuduk (25) 1953) - -

H eishansaurus pachycephalus nomen dubium ankylosau - Inst. Vert. Paleont., Peking, Chi a-yii-kuan (13) BOHLI N (BOHLIN, 1953) rid no catalogue number

Inst. Vert. Paleont., Peking, Nodosauridae gen. et sp. indet. ankylosaurid Tsondolein-Khuduk (BOHLIN, 1953) no catalogue number (17)

Syrmo saurus disparoserratus PIN 554 holotype , MALEEW(MALEEV, 1952) 1"alarllrlls disparoserralus Sheeregeen Gashoon (7) - - - (MALEEV) ZPAL MgD-I/ 115

Talarurus plicatospineus (MALEEV, PIN 557, holot ype PI N Bayn Shireh (9) 1952) 557-3

PIN, undescribed sped- - Talarurus plicatospineus Baga Tarjach (10) men, no cat alogue number - MALEEV

PIN, unde scribed sped- dinosaur ex group Ankylosaurus Bayshin Tsav (11) (SHUVALOV. 1976) men , no catalogue number

Nodosauridae gen. et sp. indet. ankylosaurs AMNH 6367 Iren Dabasu (19) (GILMORE, 1933)

Pek ing, undescribed sped- Ankylosaurs (ROZHDESTVENSKY, ankylosa urs men from Soviet-Chinese 1961) region Maortu (16) Exped.

Nodosauridae gen. et sp. indet. China ('1), no catalogue , ankylosa urs Nanhsiung (27) (CHENG el al., 1973) nu mber

Sa uroplites scutiger BoHLIN Saurop lites scutiger BOHLIN Inst. Vert. Paleont., Pe- Tebch (18) (BOHLIN, 1953) king, no catalogue number

PIN, undescribed sped- ('1)ankylosaurid (see p. 94) Khovboor (8) ankylosaur (K RAMARENKO, 1974) men. no cat alogue number --- - ankylosaurs (K URZANOV & KA- '1ankylosaurid PIN, no catalogue number Khuren Dukh (20) LA NDADZE, 1974)

und escribed sped - ankylosaurs (ROZHDESTVENSKY, ankylosaurid PIN. Kempenday (28) 1973) men, no cat alogue number

I Nodosauridae gen. et sp, indet. ankylosaur '1 Layang (26) (G ILMORE, 1933)

Peishansaurus philemys BOHLIN Inst. Vert. Paleont., Pe- nomen dubium ?ankylosaur Ehr-chiawutung (15) (BoHLIN, 1953) king, no catalogue number ANKYLOSAURIDAE FROM MONGOLIA 91

Formation and/or other Associated vertebrate fauna stratigraphic data Presumable age .--- see ROZHDESTVENSKY, 1964 1 Itemirus medullaris hadrosaurs, ornithomimids, Beleurinska svita (EFIMOV, Turonian (ROZHD ESTVENSKY, 1964) I turtles, Kansajsu chus borealis 1975) -- - possibly older than the Djadokhta Microceratops sulcidens, Chiayiisaurus lacustris, - Formation (MARYANSKA & Chiayiisuchus cingulatus, turtles OSMOLSKA, 1975) - older than the Djadokhta Forma- Microceratops gobiensis, "Stegoceras" bexelli - tion (MARYANSKA & OSMOLSKA, 1975)

Upper Cenomanian-Lower San- Microceratops gobiensis, ornithomimid indet., ha- Bayn .Sihreh svita (SHU- tonian (MARTINSON, 1975; SHU- VALOV, 1975) older drosaurs indet. VALOV, 1975) than the Djadokhta Formation (MARYAN- SKA & OSMOLSKA, 1975) . Bayn Shireh svita (BARS- Upper hadrosaurs, turtles BOLD, 1972; MARTINSON, Cenomanian-Lower San- 1973) tonian (MARTINSON, 1973)

Bayn Shireh svita (KHAND Upper Cenomanian-Lower San- & STANKEVITCH, 1975) toni an (MARTINSON, 1975) ._ - Upper Cenomanian-Lower San- Bayn Shireh or Barun tonian (MARTINSON, 1975) San- hadrosaurs, theropods, Basilemys orientalis Goyot svita see SHUVA LOV, toni an-Campanian (SHUVALOV, 1976 1976) Bactrosaurus johnsoni, Alectrosaurus olseni, Ar- Iren Dabasu Formation Cenomanian (ROZHDESTVENSKY, cheornithomimus asiaticus (GILMORE, 1933) 1977)

Cenomanian (ROZHDESTVENSKY, sauropods, hadrosaurs 1961)

Upper part of Nanhsiung Upper Cretaceous (CHENG et al., sauropods, carnosaurs, coelurids Formation (CHENG et al., 1973) 1973)

Psittacosaurus mongollensis, Prodeinodon sp. - Lower Cretaceous - Psittacosaurus mongoliensis, Changaiemys hoburen- see KALANDADZE and Ku- Lower Cretaceous: Aptian-Albian sis, Mongolemys sp. Kielantherium gobiensis and RZANOV, 1974 and SHUVA- (SHUVALOV, 1974) other not described mammals LOV, 1974

Lower Cretaceous: Neokomian Psittacosaurus sp., Iguanodon orientalis, Thoiria see SHUVALOV, 1974 and (ROZHDESTVENSKY, 1977) Aptian- namsarai, Changaiemys hoburensis NOVODVORSKA, 1974 Albian (SHUVALOV, 1974)

Sangarska svita (ROZH- Lower Cretaceous: Neocomian DESTVENSKY, 1973) (ROZHDESTVENSKY, 1973) -- ? ? Cretaceous

I ? ? Cretaceous 92 TERESA MAR YANSKA

Identification Locality (for explana- Museum cat. no tion of numbers see I original in this paper Text-fig. II I . . St egosaurides excavatus BOHLIN Inst. Vert. Paleont., Pe- nomen dubium ?ankylosaur Hui-hui-pu (14) (BOHLIN, 1953) king, no catalogue number

Lametasaurus indicus MATLEY Inomen dubium part ?an- ? (MATLEY, 1923) kylosaur

Brachypodosaurus gravis Geol. Mus. Benares Hindu CHA- I nolllen dubium ?ankylosaur Jubbulpore (22) , KRAVARTI (CHAKRAVARTI, 1934) University, V. 9 - - I ankylosaur AMNH 1959 I ! * Quoted after: BOHLIN, 1953; DASZHEVEG, 1975; EFIMOV, 1975a, 1975b ; ELZANOWSKI, 1974; GILMORE, 1933a; 1943; KIELAN-JowORowSKA, 1969, 1970, 1974b, 1975a, 1975b ; KONZHUKOVA, 1954; KURZANOV 1976; MALEEv, 1952a, 1952b, 1965b, 1956, 1955; MARYANSKA & OSM6LSKA, 1974, 1975; MLYNARSKI, 1972; Moox, 1924; NOWINSKI, 1971; OSBORN, 1924; OSM6LSKA, 1972; OSM6LSKA & RONIEWICZ , 1970, OSM6LSKA et af. , 1972; ROZHDESTVENSKY 1955, 1957, 1968; SULIMSKI, 1972, 1975; YOUNG, 1958, 1935. of Palaeontology and Stratigraphy in Ulan Bator. The collection of the Palaeontological Institute of the USSR Academy of Sciences in Moscow served as supplementary material. Some conclusions concerning the development of ankylosaurs in Asia have been based on the new material collected by the Soviet-Mongolian expeditions. The latter material is housed in the Palaeontological Institute in Moscow as will be studied by Soviet palaeontologists. Publi­ cation ofsome data on this collection is done here through kind permission ofDr. A. K. RozHDE­ STVENSKY. Abbreviations us ed for Institutions: AMNH - American Museum of Natural History, New York. BM (NH) - British Museum (Natural History), London. GI SPS - Geological Institute Section of Palaeontology and Stratigraphy, the Academy ofSciences of the Mongo- lian People's Republic, Ulan Bator, NMC - National Museum of Canada, Ottawa. NMNH - National Museum of Natural History, Washington. PIN - Palaeontological Institute of the USSR Academy of Sciences, Moscow. ZPAL - Institute of Paleobiology (Zaklad Paleobiologii) of the Polish Academy of Sciences, Warsaw. For permission to study collections I am grateful to authorities of Palaeontological Institute and Palaeontological Museum of USSR Academy of Sciences in Moscow; Dr. B. KOZHAM­ KULOVA (Laboratory ofPalaeobiology ofKazakh. Academy ofSciences), Alma Ata; Dr. A. CHA­ RIG, British Museum (Natural History), London. For help in palaeontological, stratigraphical and bibliographical matters I thank Dr. A. CHA­ RIG (British Museum, Natural History), London; Dr. W. P. COOMBS (Pratt Museum, Amherst); Dr. D. RUSSEL (National Museum, Ottawa); Dr. TH. S. PARSONS (University of Toronto, To­ ronto); Dr. M. A. SHISHKIN, Mr. W. I. ZHEGALLO, Mr. S. A. KURZANOV (Palaeontological Museum USSR Academy of Sciences, Moscow); Dr. R. BARSBOLD (Geological Institute, Academy of Sciences, Mongolian People's Republic, Ulan Bator); Dr. R. GRADZINSKI (Geolo­ gical Institute, Polish Academy of Sciences, Cracow); Dr. J. LEFELD (Geological Institute, Polish Academy of Sciences, Warsaw). 1m am especially indebted to Dr. H. OSM6LSKA (Pa­ laeozoological Institute, Polish Academy ofSciences, Warsaw) and Dr. A. K. RozHDESTVENSKY (Palaeontological Museum USSR Academy of Sciences, Moscow) for their critical remarks, and helpful suggestions and discussions. I am very grateful to Dr. W. P. COOMBS (Pratt Mu­ seum, Amherst) who critically read the manuscript and kindly corrected the English of this paper and sent me his unpublished manuscript on the Ankylosauria. ANKYLOSAURIDAE FROM MONGOLIA 93

- -- Formation and/or other Associated vertebrate fauna stratigraphic data Presumable age

sauropods, theropods ? Cretaceous

I

/

sauropods, carnosaurs Lameta Beds Upper Cretaceous

I

Finally I am extremely grateful to Dr. A. SULIMSKI {Institute Paleobiology, Polish Aca­ demy of Sciences, Warsaw) for preparing the drawings, and to Mrs. M. MALACHOWSKA-KLEIBER and Mr. L. DWORNIK (Museum of the Earth, Polish Academy of Sciences, Warsaw) and Mr. W. SKARZYNSKI {Institute of Paleobiology, Polish Academy of Sciences, Warsaw) for taking the photographs.

MATERIAL

Talarurus plicatospineus MALEEV

PIN 557 -fragments ofpostcranial skeletons of at least six individuals and two fragments of skulls. On these specimens MALEEv (1952) distinguished Talarurus plicatospineus; Bayn Shireh, Eastern Mongolia, Upper Cretaceous, Bayn Shireh svita Cenomanian - Lower Santonian according to BARSBOLD, 1972. PIN, no catalogue number -undescribed incomplete skull with cranial roof, occipital part and brain case; Bayshin Tsav, Eastern Mongolia, Upper Cretaceous, Bayn Shireh svita. PIN,no catalogue number - undescribed fragment ofmaxilla with eight teeth; Baga Tarjach, Eastern Mongolia, Upper Cretaceous, Bayn Shireh svita.

Talarurus disparoserratus (MALEEV)

PIN 554/1-2 - fragments of right maxilla and PIN 554/1-1 - left maxilla of the same spe­ cimen, described by MALEEV (1952b) as a holotype ofSyrmosaurusdisparoserratus; Sheeregeen Gashoon, Mongolia, Upper Cretaceous, Bayn Shireh svita. PIN 554/2-1 - basicranial fragment with occipital condyle, basioccipital and basisphenoid. Same "formation" and locality. ZPAL MgD-I/1l5 - right ilium. Same "formation" and locality.

Pinacosaurus grangeri GILMORE

ZPAL MgD-I1/1 - almost complete postcranial skeleton with skull of a young individual; Dayn Dzak, Mongolia, Upper Cretaceous, Djadokhta Formation, ?Santonian according to KIELAN-JAWOROWSKA, 1974a, 1974b (GRADZINSKI et al., 1968, text-fig. 29, no 5; MARYANSKA, 1971). ZPAL MgD-I1/9 - fragments of a shoulder girdle, almost complete manus, pelvic girdle, sacral 94 TERESA MARYANSKA

and caudal vertebrae, hind limb (GRADzINSKI et al., 1969, text-fig. 29, no 6). Same formation and locality. ZPAL MgD-II/31 - fragments of shoulder girdle, dorsal, sacral and caudal vertebrae, pelvic girdle, hind limbs, fragments of an armour. Same formation and locality. ZPAL MgD-II/32-fragments of caudal portion of vertebral column. Same formation and locality. ZPAL MgD-II/2 - fragments of an armour. Same formation and locality. PIN 614 - almost complete skeleton without skull, specimen described by MALEEV (1952) as a holotype of Syrmosaurus viminicaudus. Same formation and locality. PIN, no catalogue number -undescribed fragments ofskulls and postcranial skeletons ofseve­ ral individuals of various individual age (TVERDOCHLEBOV& TSYBIN, 1974). Alag Teg, Mongolia, Upper Cretaceous, unnamed "formation", stratigraphic equivalent of Djadokhta Formation.

Saichania chulsanensis gen. n., sp. n.

GI SPS 100/151 -skull and anterior part ofpostcranial skeleton: complete cervical and dorsal vertebrae, shoulder girdle, forelimb and armour in natural arrangements. Khulsan, Mongolia, Upper Cretaceous, Barun Goyot Formation, ?Middlc Campanian according to KIELAN­ JAWOROWSKA, 1974b (GRADZINSKI & JERZYKIEWICZ 1972, text-fig. 4, no 11). ZPAL MgD-I/114 -fragment of skull roof and armour. Khermeen Tsav II, Mongolia, Upper Cretaceous, Khermeen Tsav "formation" of KIELAN-JAWOROWSKA, 1975b, stratigraphic equivalent of Barun Goyot Formation. PIN, no catalogue number-undescribed, almost complete postcranial skeleton with skull. Same "formation" and locality.

Tarchia kielanae gen. n., sp. n.

ZPAL MgD-I/ll1 -incomplete skull with skull roof, occipital part and brain case. Khulsan, Mongolia, Upper Cretaceous, Barun Goyot Formation (?Middle Campanian according to KIELAN-JAWOROWSKA, 1974b) GRADZINSKI & JERZYKIEWICZ, 1972, text-fig. 4, no. 6.

"Dyoplosaurus" giganteus MALEEV

ZPAL MgD-I/43 -caudal portion of-vertebral column with tail-club. Altan Ula IV, Mongolia, Upper Cretaceous, Nemegt Formation, ?Upper Campanian - Lower Maastrichtian (GRA­ DZINSKI et al., 1968, text-fig. 4, no 5; MARYANSKA, 1970). This specimen is now housed in GI SPS in Ulan Bator. ZPAL MgD-I/42 -caudal portion of vertebral column with tail-club and fragments of armour. Same formation and locality. (GRADZINSKI et al., 1969, text-fig. 4, no 7; MARYANSKA, 1970). ZPAL MgD-I/49 - right humerus. Same formation and locality (GRADZINSKI et al., 1969, text-fig. 4, no 3). ZPAL MgD-I/113 - incomplete postcranial skeleton without skull. Altan Ula III, Mongolia, Upper Cretaceous, Nemegt Formation. PIN 551-29 - free caudal vertebrae, metacarpals and phalanges of manus, fragments of ar­ mour. This specimen has been described by MALEEV (1956) as holotype of Dyoplosaurus giganteus. Nemegt, Mongolia, Upper Cretaceous, Nemegt Formation. PIN, no catalogue number - an undescribed skeleton with skull from upper white beds of Khermeen Tsav I, Mongolia, Upper Cretaceous, stratigraphic equivalent of Nemegt Formation. ANKYLOSAURIDAE FROM MONGOLIA 95

Ankylosauridae gen. et sp. indet. (see also table I)

PIN, no catalogue number - lower j aw of an ankylosaur with preserved teeth. Vicinity of Khovboor, Mongolia, Lower Cretaceous. PIN, no catalogue number -undescribed specimen consists of humerus, dorsal and caudal vertebrae; Kempenday, U SSR, Lower Cretaceous, Sangarska svita. Palaeobiol. Lab. Kazakh. Acad. of Sci. in Alma Ata, no catalogue number - undescribed fragments of postcranial skeleton and armour. Shakh-shakh, Kazakhstan (USSR), Upper Cretaceous, Beleutinska svita (Turonian according to ROZHDESTVENSKY. 1964).

MEASUREMENTS

Table 2

Measurements of the skull (in mm)

Saichania Tarchia kielanae Pinacosaurus chulsanensis ZPAL MgD-l/lll grangeri 01 SPS 100/151 ZPAL MgD-Il/1 greatest length 455 185 greatest width (across ventral projection of squa­ mosal) 480 210 greatest width of dorsal surface of skull (just posterior to 01 bits) 395 385 203 width across center of -orbits 280 177 width at posterior edge of external nares 205 90 width between ends 01 paroccipital processes 252 ca 220 100 height of the skull (tip 01 ventral projection of squamosal to top of skull) 189 102 length trom front of orbit to front of skull 210 9S length of quadrate (condyle to squamosal contact) 160 78 greatest breadth of orbit 78 4S length of lower jaw 298 13' greatest depth of lower jaw 84 40

Table 3

Saichania chulsanensis «n SPS 100/151)

Measurements of the cervical vertebrae (in mm)

Vertebrae

C. C. C3 C. C. C. C, length of centrum 42 75 60 55 60 70 70 height of centrum anterior 52 55 50 45 50 SO posterior 60 67 68 60 52 42 43 width of centrum anterior 85 80 70 80 85 8S posterior 105 70 73 60 70 83 82 height of centrum with neural arch 80 135 160 170 150 145 150 96 TERESA MARYANSKA

Table 4 Measurements of the pectoral girdle and fore limb (in mm) T'alarurus plicatospineus Pinacosaurus grangeri Saichania chulsan ensis PIN 557-20 PIN 557-91 PIN 614 ZPAL MgD-II/1 Gl SPS 100/151 left left left left right Scapulocoracoid: total length 01scapulocoracoid 600 ca 550 580 ca 300 580 length of scapula 430 ca 380 390 ca 190 400 proximal width of scapula 122 180 90 50 75 distal width of scapula ca 150 150 ca 80 157

Humerus PIN 557-il PIN 557-91 PIN 614 ZPAL MgD-Il/l GI SPS 100/151 left left left left left length 300 320 290 ca 160 300 maximum width of prox imal end ca 135 150 135 ca 75 212 maximum width of distal end 137 150 125 ca 65 163 maximum width of shaft 53 60 51 31 73

Ulna PIN 557-80 PIN 557-91 PIN 614 Gl SPS 100/151 left left left left length 215 220 ca 210 210 maximum width of proximal end 81 120 83 125 maximum width of dist al end 67 57 48 78 maximum width of shaft 33 43 32 50

Radius PIN 557-85 PIN 557-91 PIN 614 ZPAL MgD-I1/1 Gl SPS 100/151 left left right left left length 188 171 174 182 maximum width of proximal end 77 64 62 39 84 maximum width of distal end 75 73 60 81 maximum width of shaft 37 33 30 18 41

Table 5 Measurements of metacarpal bones (in mm) Talarurus plicatospineus Pinacosaurus granger! Saich anla chulsanensis (MALEEV, 1956) PIN 557-3 ZPAL MgD-1I/9 GI SPS 100/151 Length of metacarpals: I 59 34 51 II 65 35 46 III 67 37 56 IV ca 33 58 V 49 Maximum width of proximal end of the metacarp als: I 30 22 39 II 41 23 41 III 41 22 41 IV 19 43 V 18 38 Maximum width of distal end s of the metacarp als: I 35 19 45 II 32 17 40 III 33 14 42 IV 28 35 V 28 ANKYLOSAURIDAE FROM MONGOLIA 97

T able 6

Measurements of metat arsal bones (in mm)

Talarurus plicatospineus Pinacosaurus grangeri " Dyoplosaurus" giganteus ( M ALEEV, 1956) PIN 557-3 ZPAL MgD-II/9 ( MALEEv, 1956) PIN 551 -29 Length of the metatarsals : I 29 II 81 ca 45 80 III 89 42 70 IV 87 Maximum width of the metatarsals: I 23 II 57 ca 25 70 III 40 ca 27 95 IV 47 35 Max imum wid th of the metatarsals : 1 39 24 II 67 74 III 53 22 60 IV 51 55

G ENER AL REMARKS ON SYSTEMATICS

The systema tics of Ankylosauria prop osed by Coovns (MS) is here applied. Two families are distin guished: the Ankylosauridae (including Ankylosauridae BROWN, 1908; Ankylosaurinae NOPCSA, 1923; Syrmosaur idae MALEEv, 1952) and the Nod osauridae (including among others Nod osauridae MARSH, 1890 ; Edm onton iinae RUSSELL, 1940; Acanthopholididae NOPCSA, 1902; Pan opl osau rinae NOPcsA, 1923). COOMBS' con clusion that the range of the N odosauridae is restricted to Europe and North Am erica is here confirmed. A comparison of the Asiatic and North Am erican Ankylosauridae shows th at Talarurus, Pinaeosaurus and Saiehania are less advanced in some respects (e.g. degree of development of cranial sinuses and to oth morpho­ logy) th an the North Am erican Euoplocephalus and Ankylosaurus, but some Mongolian Ankylo­ sauridae show cha racters unknown in Am erican forms. Th ese are: completely ventral orientation of th e occipital con dyle and associated special structure of th e atlas ; very stro ngly developed ventral armour and sterna l complex. Such differences mak e it necessar y to supplement COOMBS' (I. e.) diagnosis of th e Ankylosauridae. COOMBS regards as valid only two ankylosaur id genera in North Am erica: Euoplocephalus and Ankylosaurus. He regards Dyoplosaurus, Scolosaurus and Anodontosaurus as synonyms of EliOP 10 cephalus. I do not discuss th is problem, becau se I have not had an opportunity to study North American specimens. On the basis of my studies of described specimens of Talarurus plicatospineus (MALEEV, 1952a, b, 1956) and of undescribed specimens of th is genus in the PIN collectio ns I con sider Talarurus as a valid genus not congeneric with Pinacosaurus as suggested by COOMBS(MS). The new genera Saichana and Tarchia described in this pap er, the generic diversity of Talarurus and Pinacosaurus and th e validity of Sauroplites (regarded by COOMBS as nomen dubiumy all demonstrate grea ter diversificati on of the Ankylo­ sauridae in Asia than in No rth Am erica. Id entificat ion of some ankylosaur species described by BOHLIN (1953) from China is difficult. Very incomplete and badly preserved mat erial of St egosaurides excavatus and Peishansaurus philemys is the main basis for th e view that th ese two species are nomina dubia. There is no certainty about their affiliation to Ankylosauria. Assignment of Sauroplites seutiger (BOHLIN, 7 - Pala eonto logi a P ol onica No. 37 98 TERESA MARYANSKA

1953) to the Ankylosauria is unquestionable. N ewly discovered sites with the ankylosaurids, or related forms, at Khovboor and (?) Khuren Dukh in M ongolia and Kempenday in USSR confirm the occurrence of this group in Asia in Early Cretaceous. A characteristic ornamen­ tation of the armour of Sauroplites scutiger allows one to distinguish it from other ankylosaurid species. Heishansaurus pachycephalus (BOHLIN, 1953) must be regarded as nomen dubium. Its teeth are larger than those of all known ankylosaurids which makes assignment to the Ankylosauridae difficult. On the basis of that character it may be supposed that Heishansaurus pachycephalus is the only representative of the N odosauridae in Asia ; but the structure of the occipital condyle suggests its affinity to the Ankylosauridae, and existence of a W-shaped folding of the cingulum on some teeth is reminiscent of Ta/arurus. Syrmosaurus (MALEEV, 1952a) needs special discussion. Its type species Syrmosaurus vi­ minicaudus is a junior synonym of Pinacosaurus grangeri GILMORE (MARYANSKA, 1971). New materials allow one to assign the second species - Syrmosaurus disparoserratus MALEEV to Ta/arurus (see also p. 99). The third representative of Syrmosaurus - Syrmosaurus sp. belongs to Psittaccosaurus (ROZHD ESTVENSKY, 1955). It follows that there is no basis for maintaining the genus Syrmosaurus or the family Syrrnosauridae. The generic affinity of Dyop/osaurus giganteus MALEEV is an open question. Coombs (MS), regarding Dyop/osaurus as a synonym of Euoplocephalus, identifies D. giganteus as Euoplocephalus giganteus. The material on which MALEEV (1956) described this species does not furnish evidence in favour of assigning it to the North American Dyoplosaurus. On the basis of the materials collected by the Polish-Mongolian expeditions (ZPAL MgD/I/43,42, 113) assignment of Dyop/osaurus giganteus to Ta/arurus, Saichania or Pinacosaurus is excluded. An almost complete skeleton of the species, with skull, collected by the Soviet-Mongolian expeditions of 1972, is housed at the PIN in Moscow, and will be descr ibed by Soviet palaeontologists. Occipital and basicranial parts of that specimens have a structure similar to Tarchia (nov.) Until a detailed study of the above mentioned PIN specimen is done the name "Dyoplosaurus" giganteus is here used. Undescribed ankylosaurs from the Early Cretaceous at Khovboor and Kempenday housed at the PIN in Moscow are tentatively assigned to the Ankylosauridae. A detailed examination will determine its assignment either to the Ankylosauridae or to a new primitive ankylosaurian family from which the proper Ankylosauridae are derived.

CHECK-LIST OF ASIAN ANKYLOSAURIA

Family Ankylosauridae BROWN, 1908 Early Cretaceous: Sauroplites scutiger BOHLIN, 1953

Unnamed and undescribed specimen from Khovboor in PIN collection. Late Cretaceous: Ta/arurus plicatospineus MALEEv, 1952 Talarurus disparsoserratus (MALEEv), 1952 Pinacosaurus grangeri GILMORE, 1933 Saichania chulsanensis gen. nov., sp. n. Tarchia kielanae gen. n. , sp. n. Species of erroneous generic assignment: "Dyoplosaurus" giganteus MALEEV, 1956 Heishansaurus pachycephalus BOHLIN, 1953 Species of doubtful validity: Peishansaurus philemys BOHLIN, 1953; Stegosaurides excavatus BOHLIN, 1953. ANKYLOSAURIDAE FROM MONGOLIA 99

DESCRIPTIONS

Suborder Ankylosauria OSBORN, 1923 Family Ankylosauridae BROWN, 1908

Diagnosis (partly based on COOMBS, MS, emended). - Ankylosaurs with broad skulIs, width nearly equal to or surpassing length more or less triangular in dorsal view; back wide and squarely truncated; dorsal surface of skull covered by a large number of grooves which show little bilateral symmetry and which remain distinct in large individuals. Posterolateral corners of skull and nasal region with coossified dermal plates, the former tend to protrude forming horn-like, pyramidal projections; greatest width of skulI either across the tips of the posterolateral "horns" or across the distal ends of the jugal projections; nostrils variable, elongate or suboval slits openings facing strongly anteriorly (Pinacosaurus, Saichania, Euoplo­ cephalus) or small circular openings facing laterally (Ankylosaurus); nostrils divided by a septum which separates the true respiratory passage from an opening into an accessory cranial sinus; respiratory passage within skull associated with paired sinuses; occipital condyle fitted closely against the brain case, and directed postero-ventrally or ventralIy; lateral temporal fenestra and quadrate cotylus hidden in lateral view by a postero-ventral projection of the jugal and coossified dermal plate; quadrate cotylus well separated from the quadratojugal; lateral ex­ tremities of paroccipital processes do not extend beyond the posterior margin of the skuII roof; epipterygoid occasionalIy present; ectopterygoid and adjacent mandibular ramus of pterygoid directed laterally; teeth small, basal cingulum smalI or absent, but base of tooth swolIen; distal caudal vertebrae with elongate interlocking zygapophyses and long interlocking haemal arches; large terminal tail-club of dermal armour, often associated with a mass of ossified tendons; scapular spine located on extreme anterior edge of scapular blade; coracoid smalI relatively to scapula; capitulum of humerus ovate, and very slightly elevated above the shaft; ischium nearly straight or with slight anterior curvature; fourth trochanter distal to femoral mid-length; keeled armour plates usually deeply excavated on medial surface; large conical spines rare or absent.

Genus Talarurus MALEEv, 1952

Type species: Talarurus plicatospineus MALEEV, 1952a. Distribution: Upper Cretaceous of Asia, Bayn Shireh "formation" (= svita). Included species: type species and Talarurus disparoserratus (MALEEV, 1952a). Diagnosis (partly based on MALEEv, 1953, emended). - Ankylosaurid up to 5-6 m long, skulI smaII, relatively long and narrow, parietal posteriorly incurved along the medial line making occipital condyle and supraoccipital partly visible in dorsal view; articulating surface of occipital condyle variously developed, always directed ventro-posteriorly; exoccipital high, perpendicular to the skull roof; maxillary teeth with more or less developed cingulum-like thicken­ ing always cut by W-shaped furrows on external side; curvature oftooth row small; horizontal palatal shelf in posterior position; weakly developed horizontal maxillary shelf bordering internal nares anteriorly; separate cranial openings for exists or nerves situated posteriorly to foramen ovale; trunk relatively weakly dorso-ventrally flattened; manus pentadactyl, pes tetradactyl. Remarks. - COOMBS' (MS) suggestion to regard Talarurus as synonym of Euoplocephalus is unacceptable because of essential anatomic differences and different proportion of the two genera. Talarurus differs from Euoplocephalus in different shape of the skulI, in the structure of the palatal region and in having a tetradactyl pes. The reasons for assignment of Syrmosaurus disparoserratus to Talarurus are given on p. 101. In addition to Talarurus plicatospineus and T. disparoserratus, the only form showing presence 7" 100 TERESA MARYANSKA of W-shap ed folding of cingulum on maxillary teeth is Heislzansaurus pachycephalus BOHLrN (1953); however, material of H. paclzycephalus is too fragmentary to defend assignment to Talarurus.

Talarurus plicatospineus M ALEEV, 1952 (pI. 26, fig. 2; text-fig. 11A)

1952 Talarurus plic atospineus sp .n .; MALEEV, p. 273, text-figs I, 2 and 3. 1956 Talarurus plicatospineus MALEEV; MALEEV, p. 56, text-figs 1-32. Holotype: fra gmentary sku ll with pos terior part of sku ll roof, occipital regio n and basicranium. PIN 557-91, M ALEEv, 1952b, p. 273; MALEEV, 1956, p. 56, figs 1- 3. Type horizon: U pper Cre taceous, Bayn Shi reh " for ma tio n" (= svita). Type locality : Bayn Shi reh, Eastern Mongolia.

Revised diagnosis. - Talarurus with occipital condyle ventro -posteriorly oriented, visible in both lateral and dorsal views; condyle articulating surface narrow, crescent- shaped ; cingulum­ -like swelling of maxillary teeth bett er developed lingually than lab ially; maxillary teeth with weakly marked W-shaped folding of cingulum lab ially; pectoral glenoi d deep, short ; humeral head situated dorso-term inally; cha racteristic furrow-rib ornam entation of armo ur elements ; tail-club weakly developed. Material. - see p. 93. Measurements. - see tabl es 4, 5 and 6. Remarks. - M ALEEV (1956, p. 56) designat ed "a fragment of the skull (posterior part of the skull ro of, occipital region , basicran ium) and postcranial skeleton" as the type specimen of Talarurus plicatospineus (pIN 557). Examination of the rich bone mat erial belonging to this species in the PIN collections shows th at the specimen chosen as the type specimen of T. pli­ catospineus by M ALEEV (1956) consists, in fact, of bones belonging to at least three individuals. Sorting of the bones into pr oper skeleto n is imp ossible at the moment , so it seems appropri ate to regard as the ho!otype only the skull fragment described and illustrat ed by M ALEEV (1952b, p. 273 ; 1956, p. 56, figs 1-3, PIN 557-91). Th e shape of the articular surface and the relat ively stro ng backward shift of th e occipital condyle distinguish Talarurus plicatospineus from ?T. disparoserratus and from all other Upper Cretaceous ankylosaurids of Asia an d North America . Detailed compari son of all skeletal elements of T. plicatospineus with other ankylosa urids is pr esent ed in the second part of this paper. Distribution. - See tabl e 1.

Talarurus disparoserratus ( MALEEV), 1952 (pI. 26, fig. I)

1952 Syrmosaurus disparoserratus sp .n.; MALEEV, p. 134. 1954 Syrmosaurus disparoserratus MALEEV; MALEEV, p. 162, fig. 15. Holotype : two fragments of right maxi lla (PIN 554/ 1-2) and left maxi lla (PIN 554/1-1) of one in divid ual. Type horizon : U pper Cre taceous, Bayn Shireh "formation" (= svita). Type locality : Sheergeen Gashoon, Southern Mongolia. Revised diagnosis. - Talarurus with subspherical occipital condyle oriented ventro-poste­ riorly; maxillary teeth with strongly develop ed W-shaped folding of cingulum labially. Material. - see p. 93. Remarks. -MALEEV (195 2a, fig. 15) erro neo usly iden tified maxillary fragment as belonging to mandibles. His fig. 15 (1954) sho ws not a right mandible but a left maxilla. Despite the paucity of mat erial, th e fragments of maxilla and basicranium allow some ANKYLOSAURIDAE FROM MONGOLIA 101 conclusions about the development of a secondary palate and the pattern of cranial nerve openings. The labial side of the maxilla is low along the whole tooth row with a slight posterior rise toward the palatinum. Therefore along almost the whole length of tooth row the palatal vacuities are laterally bordered by a vertical core of maxilla. The horizontal palatal shelf of the maxilla is very small. The poorly developed posterior shelf is opposite the posterior third of the tooth row. Lack of an anterior horizontal shelf of the maxilla proves that the secondary palate was weakly developed and the pal atal vacuities were very large. Assignment of this species to Talarurus is tentative because of incompleteness of the material and is based mainly on similarities in general basicranial pattern, on maxillary structure and on presence of folding of the cingulum. T. disparoserratus differs from T. plicatospineus in stronger cingulum devel opment on maxillary teeth, in more distinct labial folding of the cin­ gulum, and in shape and position of the articular surface of the occipital condyle. Considering Syrmosaurus disparoserratus as a synonym of Pinacosaurus grangeri as suggested by COOMBS (MS) seems completely unjustified. Syrmosaurus differs from Pinacosaurus in more primitive structure of the pal ate, different structure of the teeth, and different distribution of cranial nerve openings. The two genera occur in formations of different age. As Talarurus is the only ankylosaurid genus in Mongolia in which the nerves from IX to XII possess separate openings and this is the case also in T. disparoserratus (PIN 554/2-1), the assignment of S. disparoserratus to Talarurus seems to be justified. Distribution. - See table 1.

Genus Pinacosaurus GILMORE, 1933

Type species: Pinacosaurus grangeri GILMORE. Genus monotypic, diagnosis and distribution as for the type species .

Pinacosaurus grangeri GILMORE, 1933 (Pls, 19-26, text-figs 2, 3 and lIB)

1933 Pinacosaurus grangeri n.sp.; GILMORE, p. 3, figs 1-3. 1935 Pinacosaurus ninghsiensis n.sp.; YOUNG, p. 5, pls 1-3. 1952 Syrmosaurus viminicaudus n.sp.; MALEEY, p. 137, figs 1-3. 1954 Syrmosaurus viminicaudus MALEEY; MALEEY, p. 147, figs 1-4, 6, 10, 12, 13 (non figs I, 5, 7-9, 11). 1971 Pinacosaurus grangeri GILMORE ; MARYA NSKA, p. 45, text-fig . I, pIs 6 and 7. Holotype: AMNH 6523, crushed skull and jaws, atlas and axis vertebrae, and a few dermal bones. Type horizon : Upper Cretaceous, Djadokhta Formation. Type locality: Bayn Dzak (Shabarkh Usu), Gobi Desert, Mongolia.

Revised diagnosis. - Ankylosaur attaining 5 m in length ; nostrils situated anteriorly, suboval, separated by horizontal premaxillary septum; septum separates a dorsal foramen leading to the respiratory passage, from a ventral foramen leading to the premaxillary sinus; young in­ dividuals show third foramen in nasal area situated laterally and leading to vast premaxillary sinus; premaxilla not completely cevered by nasal or accessory dermal "plate", without ornamen­ tation; subsphaerical occipital condyle oriented ventro-posteriorly; exoccipital slightly oblique to skull roof, causing occipital part of skull to be on same level as posterior margin of skull roof; quadrate not coossified with paroccipital process, slightly oblique ateriorly; mandibular cotylus of quadrate situated at level of posterior margin of orbit; width of premaxillary beak larger than distance between posteriormost maxillary teeth; in palatal region both anterior and posterior horizontal maxillary shelves poorly developed; palatinum relatively strongly de­ veloped anteriorly; maxillary and dentary teeth with labial cingulum and slight basal swelling lingually; orbits not completely closed; postcranial skeleton rather light, limb bones slender, manus pentadactyl, pes tetradactyI. 102 TERESA MARYANSKA

ppf

10cm ANKYLOSAURIDAE FROM MONGOLIA 103

Material. - See p. 93-94. Measurements. -See tables 2, 4, 5 and 6. Remarks. - Pinocosaurus grangeri differs from all other ankylosaurids in postcranial structure, particularly in slenderness of the limbs. The description of Syrmosaurus viminicaudus as presented by MALEEV (1954) has been partly illustrated by drawings of bones belonging in fact to Talarurus plicatospineus: fig. 7 of MALEEV (l.c.) shows a scapula of Talarurus plicato­ spineus (PIN 557-20/1); fig. 8 -a humerus of the same species (pIN 557-97f1) but diminished 1/2, not 1/3 as stated in the caption; fig. 9-radius (PIN 557-81f1) turned at 180°and ulna (PIN 557-101fI) of the sam e species. The remaining figures show bones of S. viminicaudus and only fig. I illustra­ ting a tooth is incorrect, not resembling the original. The above errors pertain only to illustra­ tions. MALEEV'S descriptions relate to bones of S. viminicaudus housed in the collections of the PIN. For a detailed comparison of the bones of postcranial skeleton see the section"Anatomy" of thi s paper. A comparison of skulls ofP. grangeri with th ose of Talarurus plicatospineus and T. disparo­ serratus reveals that the skull of P. grangeri is relatively shorter and broader, the bony palate is much stronger, teeth do not show a folding of cingulum, the occipital region does not protrude beyond the skull roof, and the openings of nerves IX- XII show a tendency to fusion. The study of the skull of a young individual of Pinacosaurus grangeri (ZPAL MgD-II/I) which is unusually well preserved, allowed me to recognize th e pre sence of cranial structures previously unknown in Ankylosauria. These are discu ssed in detail in the section "Anatomy". Distribution. -See table 1.

Genus Saichania nov.

Type species: Saichania chulsanensis sp.n, Derivation of the name: after Mongolian "saichan" - beautiful. Diagnosis. - Genus monotypic, diagnosis and distribution as for type species.

Saichania chulsanensis sp. n. (pIs 28-36, text-figs 4-8 and 11C)

Holotype: Skull with jaws and complete anterior part of postcranial skeleton, GI SPS 100/151. Type horizon: Upper Cretaceous, Barun Goyot Formation. Type locality: Khulsan, Nemegt Basin, Gobi Desert, Mongolia. Derivation of the name: after the type locality.

Diagnosis. - Ankylosaurid attaining 7 m in length; large, oval external nostrils situated terminally, divided by horizontal septum; septum separates large, suboval, dorsally situated foramen of true air passage from ventro-medially positioned passage leading to premaxillary sinus; premaxillary part of the snout relatively narrow; premaxillae partly covered by well developed secondary dermal plates; occipital condyle weakly convex, directed ventrally; epipterygoid present; exoccipital low, its dorsal part perpendicular to skull roof, its ventral

Fig. 2 Pinacosaurus grangeri GILMORE. The skuil (ZPAL MgD-II/I) A l - in dorsal view, A. - in palatal view. Abbreviations: bo - basioccipital, bs - basisphenoid, d - secondary dermal bone, ec - ectopterygoid, eo - exoccipital. cp - epipte­ rygoid, f- frontal, fo - fenestra ovalis, ic - internal carotid foramen, j - jugal, 1- lacrimal, Is -Iaterosphenoid, m - maxilla, mt - maxilloturbinal, n - nasal, oj - supposed opening of Jacobson's organ, or - orbitosphenoid, p - parietal, pa - exit of palatine artery, pf- prefrontal, pl- palatine, pm - premaxilla, po - postorbital, pof­ postfrontal, pos - postocular shelf, ppf- posterior palatal foramen, pr - prootic, prs - presphenoid, pI - pte­ rygoid ,pv - palatal vacuity, q - quadr ate, qj - quadratojugal, so - supraorbital, soc - supraoccipital, I - "tabular" v - vomer, vf- vascular foramen, vs - venal sinus, I-XU exits of cranial nerves. 104 TERESA MARYANSKA

10cm ANKYLOSAURIDAE FROM MONGOLIA 105 part strongly deflected anteriorly; quadrate oblique, with mandibular cotylus at level of middle part of orbit; orbits an teriorly and posteriorl y closed by partly neomorphic bones; skull roof overhanging occipital reg ion ; palat al region with stro ngly developed anterior and posterior maxill ary shelves ; mai n body of maxilla sur ro unds palat al vacuities over small area laterally; width of premaxillary beak almos t equa l to distanc e between posteriormost maxillary teeth; on e open ing for nerves IX-XII; atlas and axis fused; stron gly develop ed secondary plate-like intercostal ossificatio ns along lat ero-ventral part of th e trunk; limb bones very massive, fore limb stro ngly flexed, manus pent adactyl. Material. -See p. 94. Measurements. -See tables 2-6. Remarks. - Saichania chulsanensis sp. n. differs from Pinacosaurus granger! by a more advanced development of accessory cranial sinuses and a much stro nger ossification of the palatal region. In th ese characters it differs also from Talarurus. Other differences from Talarurus are: different structure of the occipital region and presenc e (in Saichania) of a single foramen for nerves IX- XII situated po sterior to the foramen ovale. The horizontal septum dividing th e external nasal opening is depressed in Saichania, but in Pinacosaurus grangeri it is situated more externally. Essenti al differences are in the occipital region. Ventral parts of the paroccipital processes of Pinacosaurus are deflected slightly poste­ riorly from the skull ro of, but in Saichania chulsanensis these processes are oriented strongly anteriorly. The occipital condyle of Saichania is stro ngly pressed into the brain case without any trace of a neck, and faces completely ventrally, invisible in lat eral and dor sal views. Well developed bon es'in the orbita l region almost completely close the orbit of Saichania. The new speci es also differs from Pinacosaurus grangeri in th e stro ng anterior tilt of the qu adrate and its ossification with th e paroccipital pr ocess. Pectoral girdle and fore limbs of Saichania chulsa­ nensis are the most massive and its metacarpals are the sho rtest of any Asian ankylosaurid. Unusually stron g development of fusion betwe en pectoral girdle and first dorsal rib, deve­ lopment of additional ossificati ons in the sternal complex and intercostal plates, coossification of atlas and axis and fusion of the quadrat e and paroccipital process distinguish the new species from all th e known ankylosaurids. Saichania chulsanensis differs from the North .American Euoplocephalus tutus in th e shape and structure of th e nostrils, and in particular in the position of the acce ssory nasal foram en. Thi s foram en in E. tutus is situa ted laterally, leads to a maxillary sinus, and is separated from the true nar es by a vertical septum. An essential difference is in th e po sition of nerve openin gs and in the occipita l condyle, which in E. tutus is directed ventro­ posteriorly. Coossification of the qu adrate with th e paroccipital process and of the atlas with the axis observed in Saichania chulsanensis have not been previou sly reported in the Ankylosauridae. In thi s respect Saichania resembles some represent atives of the Nodosauridae. Fusion of quadrat e to paroccipital process is common amo ng nod osaurids, but fusion of the atla s and axis has been reported only in Panoplosaurus and Edmontonia (a synonym ofPanoplosaurus according to COOMBS, MS). Distribution. -See tabl e 1.

G enu s Tarchia nov.

Type species : Tarchia kielanae sp. n. Derivation of the name: Mongolian "tarchi" - brain, because of a relatively large brain case. Diagnosis. -G enus mo notypic, diagnosis and distribution as for the typ e specres.

Fig. 3

Pinacosaurus grangeri GILMORE. The skull (ZPAL MgD-II/I), A1 - occipital view, A2 - lateral view, As - anterior view. Abbreviations as in fig. 2. 106 TER ESA MARYANSKA

10 ern

Fig. 4

Saichania chulsanensis gen.n., sp.n . The skull, holotype specimen (Gl SPS rOO/lSI ). A l - dorsal view, A 2 - occipital view. Abbreviations as in fig. 2. · ANKYLOSAURIDAE FROM MONGOLIA 107

10cm

pos

Fig. 5

Saichania chulsanensis gen.n ., sp.n. The skull, holotype specimen (GI SPS 100/151) in : Ai - palatal view, A2 -lateral view. Abbreviations as in fig. 2. 108 TERESA MARYANSKA

10cm

F ig. 6. Saichania chulsanensis gen.n., sp. n, The skull, holotype specimen (GJ SPS 100/15 I) in anterior view. Abbreviations as in fig. 2.

Tarchia kielanae sp. n. (p I. 27) Holotype : posterior pa rt of skull with ro of, bra in case a nd occipital region, ZP AL MgD-I/111. Type horizon : Up per Cre taceous, Ba ru n Goyot Formati on . Derivation of the name: in honour of Prof. ZOFlA K IELA N-JAWOROWSKA in rec ognition of her work on Mongolian vertebrates . Diagnosis. - Large ankylosaurid; orbits not comple tely closed ; exoccipital high and short, perpendicular to the skull roof; occipita l condyle directed ventro-posteriorly; foramen magnum higher than wide ; brain case very high; occipita l part ofskull and occipital condyle partly visible in dorsal view; one openings for nerves situated posterior to foramen ovale. Material. - See p. 94. Measurements - See table 2. Remarks. - Th e only specimen of this species is an incomplete skull which makes detailed . comparisons to oth er ankylosaurs impossible. Saichania chulsanensis has a skull almost the same size as Tarchia kielanae, but the latt er has a larger brain case, almost twice the height of S. chulsanensis. The width of the occipital part in both skulls is the same, but the position and size of the paroccipital pr ocesses and the shape and size of the foramen magnum differ considerably. The essential difference is that the occipital part in Saichania chulsanensis is obscured by the overhanging skull roof, wherea s in Tarchia kielanae the occiput is shifted slightly backwards beyond the skull roof as in Talarurus plicatospineus. Unu sually thick skull bones and almost invisible cranial sutures of Tarchia kielanae pr ove that it is an adult skull. Consequently one can tentatively state that the bony wall of the orbits and the nasal septum were not so strongly ossified as in Saichania chulsanensis. In the development of these ossi­ fications Tarchia resembles Pinacosaurus grangeri. Tarchia kielanae has some similarities to other Mong olian ankylosaurids. However, the mixture of these characters in the skull of Tarchia kielanae and the characters of the brain case prevent assignment of this species to any known Mongolian genus. Distribution. - See table 1.

ANATOMY

A det ailed description of the skull and postcranial skeleton of the ankylosaurs, based mostly on the material from North America, has been presented by COOMBS (MS). To avoid repetitions in osteological descripti ons only those skeletal elements are discussed here that are somewhat ANKYLOSAURIDAE FROM MONGOLI A 109 different in th e Asiatic ankylosaurids or could be studied in detail for th e first tim e. As the genera Pinacosaurus, Saichania and Tarchia are mon otypic , and two species of Talarurus (T. plicato­ spineus and T. disparoserratus) do not differ in gross an atomy, for the sake of brevity in anatomical descriptions I use only th e generic nam es.

OSTEOLOGY OF THE SKULL

In general the sha pes of Talarurus, Pinacosaurus and Saichania skulls have the chara cteristics of ankylo saurid skull s given by COOMBS (MS ). H orn-like protuberances on squamosals and jugals in Saichania are well pronounced and thus resembl e Ankylosaurus mor e th an Euoplo­ cephalus. Squ am osal " ho rns " are less pron ounced in Talarurus which has relatively the longest skull of Asiatic Ankylosauridae. The skull of a young ind ividu al of Pinacosaurus (ZPALMgD-II/l) ha s secondary elements of dermal origin covering the ju gals, squamosals and the region of the extern al nares that mu st appear early in development. Skull roof orname ntatio n of Saichania and Tarchia differs from that of Euoplocephalus, Ankylosaurus and Talarurus. Skull s of Saichania and Tarchia are dorsally covered by numerou s grooves, the areas between th em are smooth and elevated. Bilat eral symmetry of orname nta tion is distinctly marked. As all th e studied species are represented by single specimens, thin sections of skull roof bones have not been done. Thus I do not discuss the probl em of th e charact er of thickenin g of the skull roof (see COOMBS, MS) . Ext ern al nares of both Pinacosaurus and Saichania are visible both in ventral and dor sal views of the skull (pI. 19, fig. 1c ; pI. 20, fig. I b; pI. 28, fig. I a and 1b), and in thi s respect th ey differ from Euoplocephalus (in which the external na res are visible onl y in dorsal view) and from Ankylosaurus (in which th ey are visibl e only in ventral view).

Individual bones

Supraoccipital. The contacts of the supraoccipital are visible only in Pinacosaurus (ZPAL MgD-II/l) (pI. 19,fig. I b; text-fi g. 3).The small supra occipital connects suturally with exoccipit als, and reaches the dor sal margin of th e foram en magnum for a sho rt distance. It articulates with the parietal in Pinacosaurus. In other Mongolian ankylosaurids the supraoccipital and the pa­ rietal are coossified. Inside th e brain case the supraoccipita l contributes to th e wall of the inner ear cavity ju st above th e suture with th e exoccipita I. Exoccipital. The exoccipitals are differently developed in the several genera of Mongolian ankylosaurids. In Pinacosaurus (ZPAL MgD-II/l ) exoccipita ls are rather high (text-fig. 3), compressed antero-po sterio rly, and loosely joined to th e quadrate and squa mosal. They slant obl iqu ely backward to the skull roof and therefore are visible in a dorsal view of th e skull. Similar oblique orientat ion of th e exoccipitals is characteristic of Talarurus (PIN 551). In Ta­ larurus the exoccipital is completely fused to th e squamosal. In Saichania (text-fig. 4) the exocci­ pitals are relatively long and low, an tero-posterio rly flattened in th e dorsal part, and per­ pendicular to the skull roof. Vent rally th ey are stro ngly bent for ward. Th e paroccipital process in Saichania is completely fused to the squa mosal. In Tarchia th e exoccipitals are rather high and short, and completely fused to the skull roof and at least in part dor sally fused to the quadrat es. The ro le of th e exoccipita l in the struct ure of the occipita l cond yle and for am en magnum is distin ct in Pinacosaurus. Only the dorso-Iateral margins of th e occipita l condyle are form ed by exoccipita ls, which mak e there two sma ll protrus ions distinctly sepa ra ted from the basioccipital artic ular surface of th e occipit al cond yle. In Pinacosaurus the exoccipitals form less than a quarter of the condyle. Th e entire lateral and part of th e upper bord er of the foramen magnum are mad e up by the exoccipitals, In the lat er od orsal part of th e foramen magnum the exoccipitals in Pinacosaurus an d Saichania form two nodes pointing medio­ posteriorly with distin ct mu scular scars. It seems that the contribution of the exoccipita l to the structure of the occipital condyle is insignificant in Saichania. In Pinacosaurus the exoccipital 110 TER ESA MA RYAN SKA is perforat ed by the exit for nerve XII just above the basioccipital contact. The exit for the nerves IX -XI is situated antero-ventrally to the exit of nerve XII and almost at the suture with the basioccipital. The exoccipital form s the postero-ventral margin of the fenestra ovalis. Ins ide the brain case the exoccipital form s a part of the posterior wall of the inner ear cavity, and cont acts the opisthotic. A vast cont act of the antero-medial part of the exoccipital with the pro otic is free in Pinacosaurus. Contacts of the exoccipital with surrounding bones are obscured in other Mongolian an­ kylosaurids. There is an importa nt difference in the number of nerve exits situated posterior to the fenestra ovalis. In Pinacosaurus th ere are two openings, one probably for nerves IX -XI, and another one for nerve XII. Inside th e brain case are two corresponding foramina for the entrance of these nerves. In Talarurus four separate exits are present posteriorly and four corresponding separate entrances inside th e brain case. In Saichania (text-fig. 7) and Tarchia (pI. 27) one foramen exists for exits of the nerves IX- XII situated posterior to fenestra ovalis. Within the brain case arc two entrances corresponding to the single out er foramen. Basioccipital. A massive basioccipital form s most of th e occipital condyl e. Contact of the basioccipital with the basisphenoid is compl etely fused in the young specimen of Pinacosaurus as in all other ankylosaurids, but lat eral sutures with the exoccipitals ar e clearly marked, The ventral surface of the basioccipital is transversally concave. Along a short distance the basioccipital contributes to the ventral margin of th e foramen magnum. In Talarurus the narrow, crescent-shaped articular surface of the occipital condyle is oriented posteriorly, whereas in Pinacosaurus (pI. 19) the surface is oval and oriented postero-ventrally. In Saichania (pI. 28, fig. 1b) the articular surface is large, almost flat, and directed ventrally, and in Tarchia it is strongly convex and directed postero-ventrally. The foramen magnum in Talarurus, Pinaco­ saurus and Saichania is oval, long axis horizontal, whereas in Tarchia the height exceeds the width. Th ere is a smali unrecognized foramen at th e cent re of the concave ventral surface of the basioccipital in a youn g specimen of Pinacosaurus (ZPAL MgD -II{I). In Saichania (01 SPS 100/151) a small distinctive pouch is present in the same position, whereas in the old specimen of Tarchia (ZPAL MgD-I{lll) a large gap leading to the endocranial cavity is observed. It is probable the dimensions of this for amen depend on individual age and increase through resorption of the bone. Opisthotic. This small thin bone is partially visible in Pinacosaurus (ZPAL MgD-II{I). Tracing of its contacts cannot be done precisely because of dam age. It lines the exoccipital in the region of inner ear cavity forming part of its posterior wall. It has a short contact dorsally with the supraoccipital. The opisthotic forms a very thin bar over the fenestra ovalis. Anteriorly it contacts the basioccipital. Prootic. The prootic is a relatively larg e, massive bone. In Pinacosaurus it contacts the exoccipital post eriorly, the laterosphenoid antero-dorsally, the basisphenoid antero-ventrally, the opi thotic ventrally and the supraoccipi tal postero-medially. All contacts are loose with the exception of the last mentioned. A large foramen for exit of nerve V occurs in the anterior part of the prootic just at the lat erosphenoid cont act. Posterior to the exit of nerve V, at the contact of the prootic and basisphenoid, there is the exit ofnerve VII. The medial surface of the prootic takes part in the structure of lateral wall of the inner ear cavity. In Sa ichania there are two separate openings interpreted herein as exits for three branches of the nerve V (text-fig. 7). The smaller dorsal foramen is for VI, and the large exit located close to the basiphenoid is for Vz and v.; The entrance for nerve V is very large. The exit of nerve VII in Saichania is located in large opening that also contains exits of nerves V2 and Va but which is distinctly separated by a bony br.dge from exit of the ramu s maxillaris and the ramus mandibul ar is. The entrance of nerve VII is in the same position as in Pinacosaurus. A contact of the prootic with the epipte­ rygoid is clearly visible in both Pinacosaurus and Saichania, being particularly distinct in the latter where the epipterygoid contacts the prootic ventral to the exit of VI just above the large exit of V2 and V2 ' ANKYLOSAURIDAE FROM MONGOLIA 111

Basisphenoid. The basispenoid is small, masive and strongly compressed antero-posteriorly. In ventral view it forms slightly thickened, coarse ly rugose basal tubera at the border with the basioccipital. It sends short, massive basipterygoid processes antero -ventrally. Contact of the pterygoid processes of the basisphenoid with the basisphenoid processes of the pterygoid is movable only in the young specimen of Pinacosaurus (text-fig. 2). In Saichania these bon es are fused. Anteriorly the basisphenoid contacts the parasphenoid. In the lateral wall of basisphenoid, below the laterosphenoid contact, there is a foramen for nerve VI, and a more ventral foramen for the carotid artery. A canal leading out of this foramen inside the basisphenoid opens in the bottom of the hypophyseal cavity. The exit of the palatine artery is situate d anterior to the exit of nerve VI. A foramen, probably for the exit of nerve III , is situated ventrally to the exit of the palatine artery. The fragmentary basisphenoid of Pinacosaurus appears to form the ventral border of the exit of the palatin e artery, but the dorsal border is formed by the orbito-presphenoidal complex which also contains the exit of nerve III. Two centrally situated furrows, probably corresponding to median palatine art eries, are visible on the ventral side of the brain case, anterior to pituary cavity, in the wall made of basisphenoid. The pituitary

.'.,. - .. 5cm ...... :' L • •••••• I

" " . p . ~ V ''''''il'' ~ . _.. .~ ~ .

. . . .

-0. . -...... - . Fig. 7 Saichania chulsanensis gen.n., sp.n. Holotype specimen (G I SPS 100/151). Diagrammatic dra wing of the right side of brain case showing the distribution of exists of cran ial nerves and arteries. Obliqu e latera-anterior view, upside down. Abbreviations as in fig. 2. cavity is relatively large and deep. The dorsum sellae in Pinacosaurus (pl. 20, fig. 3) and Tarchia is low and narr ow, and in Sa ichania is broad and high. Because of differences in the height of dorsum sellae, it is pierced by nerve VI only in Saichania. A foramen for caroti s interna is visible at the bott om, and the exit of nerve VI in the lateral wall of the pituitary fossa. A distinct perpendicular furrow situate d at the level of the exits of nerve II, above and in front of the pituitary fossa in Saichania, corresponds most probably to the opt ic chiasma. Laterosphenoid. Only the posterior part of the laterosphenoid is preserved in Pinacosaurus 112 TERESA MARYANSKA

(ZPAL MgD-II/l ). It is loosely attached t o the skull roof and to the prootic and basisphenoid, as well as to ot her bones of the brain case. A well preserved lat ero sph enoid is to be observed in Saichania an d in Tarchia. In the latt er it is completely fused to adjo ining bon es. Th e lat ero­ sphenoid consists of two par ts - a massive main body that contact s the skull roof and a lat eral wing. Th e broad contact of th e lat erosphenoid with the roof covers th e frontal-par ietal suture and extends laterally to the postocular shelf.In Pinacosaurus the contact with the postocular shelf is free, but in Saichania and Tarchia th ese bones are fused, and th e main body of the lat ero­ sphenoid forms a prolongation of the postocular shelf toward the brain case. Within the lat ero ­ sphenoid near its contact with orbito-presphenoidal complex th ere is a large exit for nerve II , and postero-vent rally to it the exit of nerve IV. Parasphe noid. The parasphenoid has been preserved only in Saich ania where it is not clearly visible. A laterally flatt ened parasphenoidal rostrum wedges between the palat al wings of the pterygoid and th e vomer, making a continuati on of the interpa lata l sep tum formed by the vomer. Orbitosphenoid and presphenoid. Thi s region is not preserved in Pinacosaurus. Existence of a complete interorbital septum in Saichania suggests the presence of these bon es. Th e orbito­ presphenoidal complex is well preserved in Tarchia where it occupies a region situated antero­ dorsal to the parasph enoid . This com plex forms a lat eral margin of the olfactory channel. The shape and po sition of th e presphenoid-orbitosph enoid complex is different in Saichania and Tarchia. In Saichania these bones are relatively low but greatly expa nded antero-posteriorly, whereas in Tarchia they are rather short and high, a differe nce corr elat ed with different brain case heights in th ese genera. In Saichania the orbi to-presphenoidal compl ex contacts the additional ossificati on of th e anterior wall of the orbit. A large foram en in th e dorsal part of the int erorbital septum probabl y pierces th e orbito-presphenoida l complex. Thi s foramen is situated antero -do rsally to the exit of nerve II and lead s to a cavity below the skull roof that opens toward th e brain case. Presum abl y th e foram en is vascular , possibly for th e ophthal­ mic arte ry or it may have housed a venous sinus. P ari etal. This bone contacts th e frontal ante riorly, th e squamosal late rally, and the " tabular" postero-Iat erally. The interp arietal sutur e is very distinct. The par ietal forms th e cent ral part of th e posterior skull roof margin an d overha ngs the occipital region. The contact of th e parietal with the supraoccipital in Pinacosaurus is movabl e (pI. 19, fig. I b; text-fig. 3). An unossified gap between the parietals and the occiput occurs on ly in Pinacosaurus . All contact s of the pari etal arc visible on the internal side of th e skull roof. In its posterior part , th e parietal forms two nodes th at fit into corresponding depression s in the supra occipital. Frontal. In Pinacosaurus a relativ ely small, almost square frontal is separated from the parietals by a transverse suture . The naso-frontal suture is very distinct and is directed trans­ verse ly (pI. 20, fig. 1b; text-fig. 2). The fron tal contacts th e prefrontal antero-medially, whereas it meets the postfrontal lat erally. Th e frontal postorbital contact is sho rt. Th ere is a distinct interfrontal suture. All contact s of the frontal are visible on the inn er side of the skull roof. Ke el-like ridges on th e inner surface of the frontals diverge anteriorly and constitute the medial extremity of th e preorbital septum. More medially, transverse keel-like ridges of th e frontals form th e anteri or limit of the brain case. The contact of the frontal with th e lat ero ­ sphenoid close to the suture with th e parietal is loose and is mar ked by nu merous fine furrows. Premaxilla and secondary dermal ossification of the narial region. Only in the Pinacosaurus specimen ZPAL MgD-II/l is the premaxilla well exposed (pI. 19, fig. 1c; pI. 20, fig. 1 and 2; text-figs 2 and 3). My former description (MARYANSKA , 197I) of the premaxilla of that specimen ne eds some corections, The pr emaxilla forms half of the preorbital par t of th e skull. Lat erally the premaxilla-maxilla suture runs vertically in its ventral part. More dor sally th e premaxilla protrudes backward as a process wedged between th e maxilla and the nasal but not reaching the lacrimal. Antero-Iaterally three pairs of external openings pierce the premaxillae above th e toothless beak (pI. 20, fig. 2). Two ventrally situa ted pairs are completely surrounded by ANKYLOSAURIDAEFROM MONGOLIA 113

premaxilla, which also forms ventral margin of the third pair of open ings. Along the midline the premaxilla has a narrow dorsal process that has a very short suture with the nasal. Dorsal and dorso-lat eral rims of the third, dorsally positioned openings are formed by secondary dermal ossifications. Th e latter contact the nasals with a distinct suture and overlap the premaxilla anteriorly. In effect, the nasals are excluded from the dorsal margin of the dorsally situat ed nasal openings. Because of these second ary ossifications the external nares open far to the front, as otherwise they would open much farther posteriorly on the dorsal side of the skull. It may be presumed that similar relations exist in all other ankylosaurids. It seems probable that the dorsal surface of the premaxilla of Ankylosaurus is secondarily hidden also by strongly developed secondary dermal ossifications of that regio n. All known ankylosaurs have a concave ventral surface on the premaxilla except a young individual of Pinacosaurus in which it is convex, probably a juvenile character. Imp ortant differences between Pinacosaurus (text-fig. 3), Saichania (text-fig. 6) and Euoplocephalus concern th e number and distribution of openings through the premaxilla in the externa l nasal region. Ank ylosaurus is not mentioned here because of the lateral position of its nares and lack of the data concerning the number and situation of the particular openings. As mentioned above, a juvenile Pinacosaurus has three pairs of openings, whereas Euoplocephalus and Saichania have but two of them . Th ese are, however, po sition ed differently in these two genera. In all three genera the opening situated dorsally leads to true dorsal air passage and the other openings lead to accessory cranial sinuses. In Euoplocephalus the additional foram en is situated laterally in relation to the dorsal foramen and leads (according to COOMBS, I. c.) to a maxillary sinus . In Saichania this second foramen is situat ed ventro­ medially and leads to a premaxillary sinus. Pinacosaurus has two foramina leading to the pre­ maxillary sinus, th e lateral one leading to that part of prem axillary sinus which is directly connected with the maxillary sinus. Th e location of th ese foramina in Pinacosaurus seems to suggest that one of them may correspon d to that of Saichania , and the other to that of Euoploce­ phalus. Closing of lateral foramen in Saichania and of the medio-ventral one in Euoplocephalus may be explained by different developm ent of accesso ry cranial sinuses in these genera. The pre­ maxillary sinus is most extensive in Saich ania, whereas in Euoplocephalus the maxillary sinus is larger. Maxilla and maxillary sinus. Maxillary contacts with surrounding bon es are visible only in Pinacosaurus (pI. 21, fig. 1). Laterally the maxilla does not reach the externa l nasal openings. The maxilla-nasal contact is distinct but short. In \2alatal view the maxilla of Pinacosaurus shows the incipient anter ior palatal shelf (text-fig. 2). A small process of the maxilla ju st at its contact with the premaxilla is situated antero-Iaterally to the palatal vacuity and does not reach the vomer, so that the anterior bord er of palatal vacuity is form ed by th e premaxilla. The posterior palatal shelf of the Pinacosaurus maxilla is weakly developed. Th e palatal shelves ofthe maxilla are very well developed in Sa ichania (text -fig. 5) as in Euoplocephalus. Development of palatal maxillary shelves form ing a secondary palate results in a decrease in participation of the main body of the maxilla in the lat eral rim of the palatal vacu ity. An int ernal maxillary protuberance in the air passage region is visible through the palatal vacuity of both Pinacosaurus and Sa ichania. Th e maxillary sinus present in all Ankylosauridae is divided by a vertical septum at approximat ely mid-length. Both parts of the sinus open to the nasal cavity. Th e bony process separat es the last eight maxillary teeth from the anterior ones in the maxilla in Saichania, creating a sort of diastema placed exactly at the level of the interna l sept um in the sinus. The "diastema" correspond s in length to about one and a half of alveola. I have mention ed a connection between maxillary and premaxillary sinuses in Sa ichania. The number of teeth in a maxillary row differs in genera of Ankylosaur idae. Th ere are 19 teeth per row in Talarurus, 17 - in the young specimen of Pinacosaurus, 22 - in Saichania, whereas 22 - in Euoplocephalus and 35 - in Ankylosaurus. Lacrimal. Th is bone is well exposed only in the youn g specimen of Pinacosaurus (ZPAL MgD-II/l). Antero-ventrally it contacts the maxilla along a suture that slants obliquely upward, 8 - Palaeontolog ia p olonica No. 37 114 TERESA MARYANSKA and anteriorly it contacts the nasal (text-fig. 3, pl. 19). The lacrim al doe s not reach the pre­ maxilla from which it is distinctly sepa rated by a maxilla-nasal cont act. Dorsally the lacrimal contacts the prefrontal and suprao rbital; postero-ventrally below the orbit it contacts the jugal. The inner wing of the lacrimal forms the antero-lateral orbital wall. The lacrim al foramen is clearly visible. Medially the inner orbital wing of the lacrimal contacts the accessory orbital ossification of the anterior orbital wall. Jugal. The jugal is well exposed in a young specimen of Pinacosaurus. It form s the ventral boundary of the orbit. Ant eriorly, it contacts the lacrim al lat erally and within the orbit (text-fig. 3). Cont act with the maxilla is lat ero-ventral, just at the maxilIa-ectopterygoid suture. Posteriorly, on the lateral side of the skull and within the orbit, the lacrim al contacts the post­ orbital. The ventro-lat eral margin of the jugal is overlapped by the quadrat ojugal. Posteriorly, a horizontal wing of the jugal protrudes into th e orbit to form the postocular shelf (Hxss, 1969). In the young specimen of Pinacosaurus the secondary dermal plate is not coossified to the jugal, but only to the quadratojugal and squa mosal. However, the jugal of adult in­ dividuals of Saichania and "Dyoplosaurus" is partly covered by that dermal plate . Quadrate. This bone is developed differently in Mongolian ankylosaurids than in the North American genera (C OOMBS, MS). The only Mongolian ankylosaurid in which the ju nction of quadrate and squamosal is streptostylic is the young individual ofPinacosaurus (ZPA L MgD-II!l). In all other representatives of the Ankylosauridae (Saichania, Tarchia, "Dyop!osaurus") the quadrate and squamosal are fused. The contact between the quadrate and the paroccipital process is free in young Pinacosaurus and in adult"Dyoplosaurus" (PIN, und escribed specimen, no catalogue number) and probably in Tarchia. In Saichania the par occipital process is firmy fused to the quadrate. The pterygoid-quadrat e junction is fused in Saichania, Tarchia and " Dyoplosaurus", and only in the young Pinacosaurus does the quadrate process of pterygoid freely overlap the pterygoid process of the quadrate. A coossification of the quadrate with the squamosal and with the paroccipital process stated in Mongolian ankylosaurids resem bles the conditions in the Nodo sauridae (CO.JMBS, I.e.). The quadrate is directed somewhat anteriorly in Pinacosaurus and strongly anteriorly in Saichania. In Tarchia and"Dyoplosaurus" it is almo st vertical and perpendicular to the skull axis. QuadratojugaI. In Pinacosaurus the quadratojugal is a small bon e with two narrow pro­ jections. The anterior horizontal one underlines the jugal ventra-medially along the posterior half of suborbital bar. The longer, posterior projection is almost vertica l and embraces about 3/4 the length of the lateral margin of the quadrate. The ventral margin of quadrat ojugal­ quadrate con tact is definitely above the man dibular cot ylus. Laterally, the vertical projection of the quadratojugal reaches the medial surface of the postorbital. Externally the quadratojugal is covered by a secondary dermal plate. In Saichania all contacts of the quadratojugal are obscured. Squamosal. The squamosal in Pinacosaurus (ZPAL MgD-II/l) is a small bone that contacts the postorbital, the parietal, the second ary dermal-squamosal, and the "tabul ar". Ventrally on the squamosal is an articular socket for the quadrate. In Pinacosaurus the squamosal-quadrate junction is streptostylic, in Saichania these bones are fused, but with distinct boundaries between them; and in Tarchia the fusion of the squamosal and the quadrate is so complete that the boundaries are untraceable. This character is probably connected with age of the ind ividuals. Postorbital. The postorbital is a relatively large bone developed in three planes. Suturally it is distinguishabl e only in Pinacosaurus (ZPAL MgD-IIfl). The horizontal wing is incorporated into the skull roof and cont acts the postfrontal and posterior supraorbital anteriorly, the frontal medially, the squamos al medioposteriorly, and the secondary dermal plate, which forms postero-lateral corner of the skull, also medio-posteriorly. Th e vertical wing of the post­ orbita l forms the posterior part of the orbit and contacts the jugal and secondary dermal­ jugal bone ventrally. The third, internal wing of the postorbital forms the postocular shelf. This structure extends dorsally from the horizontal wing of the postorbital to the brain ANKYLOSAURIDAE FROM MONGOLIA 115

case contacting both parietal and laterosphenoid. In Pinacosaurus the contact of the postocular shelf with the laterosphenoid is loose. Ventrally in the orbit the postorbital part of the postocular shelf contacts the ventral wing of the jugal, thus forming the postero-ventral wall of the orbit. In the young specimen ofPinacosaurus the postocular shelf is low; in Saichania it is more massive and more ventrally expanded. It follows from an examination of the skull of Pinacosaurus, that the ventral part of the postocular shelf is built of the jugal, but the posterior part is composed of postorbital. Contact of postorbital with the squamosal is visible on the ventral side of skull roofjust posterior to the postocular shelf. Prefrontal. The suturally distinct prefrontal of Pinacosaurus is a relatively large bone that contributes to the construction of the skull roof. Antero-medially it contacts the nasal; postero­ -medially its contact with the frontal is short. Postero-laterally the prefrontal contacts the postorbital and laterally the supraorbital. The contribution of the prefrontal to the structure of the lateral wall of the skull is insignificant. Ventrally it has an almost horizontal suture with the lacrimal. Supraorbitals. Two supraorbitals are evidently incorporated into the skull roof in Pina­ cosaurus. They form the upper orbital margin and protrude over the orbit as horns (text-fig. 2). The anterior supraorbital contacts the lacrimal, the prefrontal, and the postfrontal ; the posterior one contacts the postfrontal and the postorbital. A suture between the supraorbitals is well visible. Unlike Euoplocephalus (COOMBS 1972) in Mongolian ankylosaurids the bony eyelids do not occur. Postfrontal. The bone here called the postfrontal has been found in Pinacosaurus (ZPAL MgD-IIfl). It takes part in the structure of the skull roof and is the only skull bone that forms the roof of the orbit (text-fig. 2; pl. 20, fig. 1b). In dorsal view, it contacts medially the frontal, and anteriorly the prefrontal, posteriorly the postorbital, and laterally two supraorbitals. On the internal side of the skull all the sutures of this bone are distinct. Aside of its contacts visible on the dorsal side of the skull, the postfrontal contacts in the orbital region an additional ossification that forms the anterior wall of the orbit. The contact of the postfrontal with the frontal is situated slightly lateral to the contact between the frontal and laterosphenoid. The bone described here as the postfrontal is regarded by COOMBS (MS) as a third palpebral. Exten­ sive participation of this bone in the structure of the skull roof and the orbital roof as well as the fact that it attaches to the brain case preclude identification as a palpebral (or supraorbital). Nasal. The largest bone in the skull of Pinacosaurus (ZPAL MgD-II/l) is the nasal which reaches far backward in the skull roof (pl. 19; text-fig. 2). A transverse naso-frontal suture runs approximately opposite the mid-point of the orbit. The nasal-prefrontal contact is long, and passes from the dorsal to the lateral side of the skull. In the lateral wall of the skull the nasal contacts posteriorly the lacrimal, ventrally along a shortdistance the maxilla, and posteriorly the premaxilla. The nasal-premaxilla contact is well exposed only on the lateral wall ofthe skull. Antero-dorsally along its whole width the nasal contacts the secondary dermal nasal ossification with a distinct suture.This secondary ossification coversthe premaxilla-nasal in that region. Under this ossification in the median line of the skull, the nasal contacts the narrow ascending pre­ maxillary process. Development of an extensive secondary dermal nasal ossification excludes the true nasal from the rim of any opening in the narial region. The internasal suture is distinct and incised. Inside the nasal region along the whole length of the nasal bones, a well developed internasal bony septum is present. In Pinacosaurus it is formed presumably at least in part by the nasals. There is a massive bony plate situated medially within the nasal cavity at the internasal suture. This plate is on close contact with its fellow and forms a medial section of the internasal septum. The anterior prolongation of this septum is made by the premaxilla. As along the whole length ofthe nasal, the internasal septum is definitely paired and is formed by the internal wings of the nasal; it cannot be regarded as an ossification within septum nasi. As it may be observed on the skull of a young specimen of Pinacosaurus the septum between two air passages has nothing in common with the vomer. There is a distinct gap between the 116 TERESA MARYANSKA vertical vomeral keel extending high dorsally and the na sal septum hanging down the skull roof in Pinacosaurus. Becau se of th e immaturity of th e specime n of Pinacosaurus th ere are no ossifications within th e septum nasi to complet e the int ern asal sept um. In Saichania a complete intern asal septum is develop ed in which both nasals and other ossifications within septum nasi as well as those of th e ehtmoidal region take part. This internasal septum in Saichania attains the vome ral ke el. Pterygoid. In bo th young Pina cosaurus (ZPA L MgD-II/I) and in adult Saichania (GI SPS 100/151) contact of the basipterygoid eal tuber with th e pt erygoid is very distinct but fused. The over­ lapping contact of th e qu adratic win g of the pterygoid wit h th e qu adrat e is loose only in a young Pinacosaurus (text -fig. 3) but fused in adult Saichania (text-fig. 5). Anteri orly, thin vertical pterygoidal sheets form posteri or secti ons of th e palat al wall and in the median line of the palatal region are overlapped by the vo me r. Cont act of th e pt erygoid with the palatinum is clearl y visible in both specime ns in qu estion. It is situa ted in the vertical wall that forms the p osterior limit of the palat al region. The pterygoid exte nsively overlaps the palatinum, so that a large part of the posterior palatal wall is formed by two very thin bones. The ventral pterygoid part of this wa ll is pi erced by a n opening. In th e posterior part of a highly roofed palatal region where palat al septum is form ed of pt erygoid s, th ere are relatively large posterior palatal for­ amina, presen t in both Pina cosaurus and Saichania. They are separated in the median line by the pterygoids and the parasphenoid . The vom er forms th eir antero-medial margin whereas the antero-lateral, lat eral and poster o-Iat eral borders are formed by the palatinum, an d the postero-medial border by th e pt erygoid. Con tact of the pterygoid with the sma ll ecto pterygoid is visible in Pinacosaurus on the lat erally d irected mandibul ar processes. D istinct contact of the pterygoid with a bone here nam ed the epipte rygoid is visible in Saichania. The dorsal part of the qu adratic wing of the pterygo id is embrac ed by th e epipte rygoid just at th e base, lateral to contact with th e basisphen oid . Epipterygoid. The bon e here call ed the epipterygoid was not found in the Ankylosauria and among other Orn ithischia it occur s only in the Pachycephalosauri a (MARYANSKA & OSM6L­ SKA, 1974). It is a rod-like bone of small diam et er exte nding almo st parallel to the brain case (pI. 28, fig. 1b). It embraces a nteriorly the pterygoid with its two sma ll processes, an d rests upon th e prootic posteri orly and slightly d or sally , in front to the exit of the nerve V. T his very delicate bone is preserved completely onl y in Saichania (GI SPS 100/151) and in the PIN undescribed specime n, no catal ogue number , collect ed at Khermeen Tsav. Its presence is a result of the perfect state of preservation of th e whole skull as in Prenocephale (MARYANSKA & OSM6L­ SKA, I.c). A distinct trac e of it is observed in Pinacosaurus (ZPAL MgD-II/I). I have fou nd it also in Euoplocephalus in the specimen BM (NH) R. 8588 which is a cast of AMNH 5403. It seems prob able that it was characteri stic of all the Ankylosauridae. Ectopterygoid. This sma ll tri angular bone is distinct only in a young specimen of Pina­ cosaurus. It joins the pt erygoid with th e max illa an d reaches the palatinum on the pat atal side. Palatine. T he palatine is well visible in Pina cosaurus and in Saichania (pI. 2 1, fig. 3 and pl. 30, figs I and 2). It is a part of the highly vaulted palat e. Contact of the palatine with the vomer runs on the lateral wall of the palat e just beneath its roof, so the palatine forms lateral wall of the V-shaped palat e. The palat ine form s posterior an d medi o-posterior margins of the palat al vacu ity .D istinct sutural cont act of the palatine with the maxill a is visible postero­ laterally to the palat al vacuity. The contact with ectopte rygo id is very short. The palatine is situated lower posteri orl y th an ante rio rly. In Pinacosaurus the palatine contacts the main body of the maxilla along the last 4-5 maxill ary teeth. Con tacts of the palatine with the pterygoid is very distinct in Pina cosaurus. The palat ine underl ies the vertical wall th at di vides the palatal region posteriorly. D orso-posteriorl y the palat ine contacts the parasphenoid. In Saichania the palatine-maxilla contact is more medi ally situated and the palatine does not contact the main bo dy of the maxilla, but rather the posteri or hor izontal palatal shelf of the maxilla. This contact is broader than in Pinacosaurus and corresponds to the part of the maxilla occupied by ANKYLOSAURIDAE FROM MONGOLIA 117

8-9 last maxillary teeth . Among Mongolian ankylosaurids the horizontal palatal part of- palatine is best developed in Saichania and less in Talarurus in which the palatine contacts the main body of maxilla far posteriorly along the last 2-3 maxillary teeth. Presence of num erous pits and chamb ers in the posterior part of the palatine is characteristic of all specimens examin ed. Th ese pit structures are best develcped in Saichania (pI. 30, fig. 1). In Pinacosaurus th ey are pre sent only in the posterior part of the palatine, whereas in Saichania the entire palatal part of palatine is covered by these structures. Th e palatine form s a system of chambers and pits separated one from another by thin walls. During preparation of the palatal region it appeared that the whole palatine surface was covered with fine bone debris, so it seems probable that the pits and chamb ers may have been closed in life and for med a kinds of closed gas-filled cavit ies. Thi s pneumatization of the palatin e may be con nected with a tendency to lighten the heavy skull, or the se chambers may have been resonance boxes. Vomer. On the well preserved Mong olian specimens it is demonstrated that the vomer did not reach the skull roof. This pertains to Pinacosaurus (ZPAL MgD-II/I), Saichania (GI SPS 100/151) and Talarurus (PIN, unde scribed specimen, no cata logue number, from Baynsheen Tsav) in which the vomer and the skull region situated dorsally to vomer are preserved. The very thin and deep vertical plates of the vomer form a septum within a highly vaulted palatal region and do not pass to skull roof dorsally. Th e vomeral septum extends far ventrally. In Pinacosaurus (ZPAL MgD-ll/l) it attai ns the ventral line of tooth row of the maxilla and in Saichania the ventral margin of that septum is situated below the ventral line of the tooth row. The lateral extension of the dorsal part of the vomer is stro ngest at the contact with the pre­ maxilla. In Saichania a laterally directed wing of the vomer joins anterior palatal shelf of the maxilla, both bones formi ng the anterior margin of the palatal vacuity. Th e vomer of Pina­ cosaurus does not attain the maxilla (text-fig. 2). In the anterior part of the palatal region the vomer dorsally contacts a massive internasal septum. In the postero-dor sal part of the palatal region the vomer embraces the parasphenoid. Contact of the vomer with the pterygoid is clearly visible in Pinacosaurus and Saichania, Posteriorly, the vomers evidently overlap very thin plates of the pterygoid which form the posterior part of the palatal septum. The vomer attains the interpterygoid foram en. Mandible. Some inform ation can be added to the descripti on of the mandible ofankylosaurids given by COOMBS (MS). The main differences between the structure of the mandible of the North American and Mongolian ankyl osaurids is the position of the angul ar. In Pinacosaurus (ZPAL MgD-II/l) the angular is a small ornamented bone, the anterior process of which hardly attains the penultimat e alveolus. Small dimensions of the angular in Pinacosaurus may be attributed to the youn g age of the individual. In an adult indiv idual of Saichania the angular attains but half the length of the tooth row. This bone is definitely shorter in Mongolian ankylosaur ids than in North American genera in which it terminat es at about the level of the anteriormost alveolus. Some differences are also observed in the structure of the coronoid. Both in Pina­ cosaurus and Saichania the coronoid is a relatively large bone the anterior wing of which attains more than a thi rd of the length of the tooth row. The posterior, vertical wing of the coronoid forms a relatively high process definitely higher than the coronoid process of the surangular. The articular of Pinacosaurus is an extremely small bone limited to the glenoid. In Saichania the articular did not continue at the bottom of the mandibular fossa. D ifferences in mandibular shape in Pinacosaurusand Saichania (pl. 22, fig. 8; pI. 29, fig. 1) is probably relat ed to differences in age of the individuals. In the form er th e mandible is rath er low and equally high almost over its length (excluding symphyseal part) with a slightly marked anuglar protuberance . Seventeen functional teeth are present in the right ramus, and 16 in the left ramus of Saichania mandibles; 16 teeth are preserved in the left ramus of a Pinacosaurus mandible. Some differences may be not ed in the structur e of mandibular teeth in Pinacosaurus and Saichania. Tooth crowns of Pinacosaurus are higher relative to their width and have a weakly mark ed basal swelling on the lingual side (pI. 22, fig. 7). In Saichania this swelling takes the form of a distinct cingulum 118 TERESA MARYANSKA

which is very prominent both on the lingual and labialsides. Also the manner in which the central cusp passes into a flat broad keel on lingual side of tooth is better marked in Pinaeosaurus than in Saichania. Accessory ossification in antorbital wall. A completely ossified antorbital wall occurs in Saiehania (GI SPS 100/151). First stages of the development of this wall are observed in Pi­ naeosaurus (ZPAL MgD-II/I). An additional ossification situated perpendicular to the skull roof contacts the orbital wing of the lacrimal laterally and the prefrontal, the supraorbital and the postfrontal dorsally. Its postero-medial cont inuation is the preorbital crest on the internal surface of the front al. Probably this ossification originates within lamina orbitonasalis and in Saichania increases its size and closes the orbit in the front, or joins other additional ossification of this region. Accessory ossification in posterior border of the skull roof. In Pinaeosaurus the posterior border of the skull roof, which overhangs the occiput, contains a small bone bordered by distinct sutures. It is here designed the "tabular" (text-fig. 4, pl. 20, fig. 1). It contacts the parietal antero-medially, the squamosal ant eriorly and the secondary dermal bone laterally. The ho­ mology of this bone is uncertain as its presence has not previous been reported in the Orni­ thischia. Secondary dermal ossifications. A young individual ofPinaeosaurus has additional secondary ossifications of dermal origin which cover partly the skull bones. Two pairs of secondary bones are present in the nasal region (text-fig. 2 and 3). One of them, situated centrally in front of the nasals, partly covers the premaxilla. Anoth er pair overhangs the dorsal nasal opening from dorso-Iateral side. Another two pairs of dermal ossification occur in the posterior region of the skull. One of them is coossified with the squamosal and the postorbital and forms horn-like protuberances at the posterior skull margin. The second pair forms lateral bony projections contacting the squamosal, jugal and postorbital. Aside from these four pairs of large dermal bones, in various parts of the skull there are fine bony tub ercles that can be observed e.g. on internasal suture and near the maxilla-premaxilla suture on lateral side of the skull.

Accessory elements Auditory apparatus. A part of the left stapes is preserved in the young specimen of Pina­ eosaurus (pl. 20, fig. 4). The bone is rod-shaped, 15 mm long as preserved with a diameter of 2 mm. The expended foot plat e is not preserved. Hyoid apparatus. A part of hyoid apparatus is preserved in Saiehania (GI SPS 100/151). It consists of two bent, rod-like bones and a central, unpaired bone that is bifurcated distally and has a convex dorsal surface (text-fig. 8; pI. 29, figs 6 and 7). Fragments of thin, battered bony sheets touch distal ends of the central unpaired bone. Paired, rod-like elements of the hyoid app aratu s may be compared with tho se found in Panoplosaurus and Euoploeephalus (COOMBS, MS) and in Proto eeratops and Psittaeosaurus (COLB ERT 1945) interpreted as first ceratobran­ .chials. In Saiehania no element resembling the second ceratobranchial of Protoeeratops or Psittaeosaurus (COLBERT, I.e.) has been found . The unpaired centrally located bon e of Saichania is interpreted herein as the corpus (basihyal and basibranchial) with strongl y developed, narrow lingual process (entoglossal process). The rod-like bones represent probably the first cerato­ branchials. The fragments of the bony sheets situated near the posterior ends of the corpus may correspond to cornu hyale of FORBRINGER (1922), and ceratohyale of CAMP (1923). Some of these fragments may be derived from second ceratobranchials. If the position of the above sheets is natural, the first interpretation is more probable. It is also possible that the distal ends of the corpus may represent second ceratobranchials, but there is not the slightest trace of coossification of these elements with the corpus, which makes such an interpretation doubtful. The hyoid apparatus of Saiehania is probably the first known so completely ossified a pparatus in dinosaurs. A reconstruction of the hyoid apparatus of Saiehania shown in fig. 11 differs considerably from that given by COLBERT (1945) for Psittaeosaurus and Protoeeratops. COLBERT ANKYLOSAURIDAE FROM MONGOLIA 119

5cm

",,,,'

,'" '" " \ \ I/I '" -, "\ \ \ I' , I I "' - "l - - ' \ I I I \~ ~I \ I \I \ I \ I \ 1 \ \ '\ \ \ \\ \ \ \ \ \ \

Fig. 8 Saichania chulsanensis gen. n., sp.n , Holotype speci men (GJ SPS 100/15\). Reconstruction of the hyoid apparatus. Abbreviations: e - corpus, ell - ceratohyal, eb - ceratobranchial I, entp entoglossal process.

(I.c.) bases his reconstruction on the hyoid apparatus of recent crocodiles in which the carti­ laginous corpus is broad without well developed entoglossal process. The structur e of ossified, clements of that appara tus in Saichania clearly shows th at it is closer to th at of Sphenodon (FURBRINGER, 1922) or Lacerta (GOODRICH, 1930).

Bony nasal structures Nasal cavities of Pinacosaurus grangeri and of Saichania chulsanensis were easily emptied of loo se sandy sedim ents. Th ere are differences in the structure of nasal caviti es in the above forms, both in th e number and dim ensions of ossifications. It is difficult to say whether differences are correlated with different ontogenic stages or are taxonomic vari abl es, becau se only one specimen of a young Pinacosaurus (ZPAL MgD-I1/\) and an adult Saichania (GI SPS 100/151). were at my disposal. In Pinacosaurus a well developed bony ridge is present along the whole length of the intern al sur face of th e nasal in the dorso-lat eral part of the nasal cavity. Similar pattern of longitudinal ridges along the und erside of th e nasals is found in som e Theriodontia, e.g. cyn od onts (Diademodon, Nythosaurus) , gorgon op sids (Leontocephalus) , wha itsioids (Moschowhaitsia) an d tritilod on ts (Oligokyphus) (WATSON, 1913; KEMP, 1969; TATARINOV, 1963, 1974; KOHNE, 1965). In theriod onts such ridges are compared to th e bony ones occurring in mamm als (crista nasoturbinalis), and their presence is taken as evide nce for existence of nasoturbinals. The same inte rpretation can be applied to th e ridges on the na sals of Pinacosaurus. In th e anterior part of nasal cavity in Pinacosaurus the nasal ridge joins a massive bony protub­ erance on the level of the lat eral wing of th e nasal, lacrimal, and maxilla (text-fig. 2) slightly posterior to th e lat eral premaxilla-maxilla suture. This conchal protuberance is convex antero­ medially, and concave postero-lat erally. Its third wall is bowl-like, and ventro-posteriorly 120 TERESA MARYAN SKA oblique. Along the skull roof this ossification forms a sharp, posteriorly dir ected ridge that passes into the nasal ridge described above . This ossification was described in Pin acosaurus (MARfANSKA, 1971 ) and interpreted as a maxilloturbinal. Th e whole bony structure in question is clearly visible both within the external nares and through the palata l vacuity. A parti al bony internasal septum is another characteristic featur e of the nasa l cavity of Pina cosaurus. Th e anterior part of this septum is formed by a median ascending process of the premaxilla and the posterior part by massive, vertical wings that arise media lly along the whole length of the nasal. Th ese wings adhere closely one to another in th e median line of the skull roo f. Th ey are high ante­ riorly, and diminish gradualy toward the rear. In fossil reptiles kno wn to me the nasal septum is never formed by the nasals. In some theriodonts only few ridges are formed by them. Saichania chulsanensis. Th e nasal cavity of Saichania is very larg e and occupies the entire preorbital length of the skull (text-fig. 9). Three bony structures (crista naso-turbinalis, maxillo­ turbinal, and nasal septum) pre sent in Pinacosaurus attain a much higher degree of the ossifica-

5 em

Fig. 9 Sa ichania chulsanensis gen.n., sp.n . Holotype specimen (GJ SPS 100/151). Ossified conchaI system in palatal view. Part of the right maxilla, pterygoid and palatine removed. Ab breviations: IIlI - maxilloturbina l, III - nasot urbin al, elhl - ethmot urbinal region, icr - interconchal recess; other abbreviations as in fig. 2.

tion in Saichania. In add ition Saichania has bony lamellae situated postero-dorsally in the nasal cavity. The nasal cavity of Saichania is completely divided into two pairs (right and left) by massive bony septum. Ant eriorly, the int ernasal septum is composed of premaxilla and nasal and attains the vomeral keel ventrally. The parasphenoidal rostrum and an additional ossification ANKYLOSAURIDAE FROM MONGOLIA 121 within the septum nasi are located ant ero-dorsally and take part in formation of internasal septum posteriorly. Thi s ossification is here identified as an ethmoidal ossification. It extends anteriorly up to the part of the septum made by the nasal. Posteriorly the nasal cavity is closed by ossifications within the lamin a orbitonasa lis which forms the anterior orbital wall. The posterior part of the nasal cavity lies very close to the telencephalon ; hence the olfactory lobes were very short. BROWN and S CHLAIKJER (1943) pointed out the existenc e of very short olfactory lobes in ankylosaa rs and not ed their compl ete coverage by ossi­ fications of the frontal part of the brain case. The description given below is based on the structure of the right half of the nasal cavity which is better preserved. The external nasal opening of Saichania is horizontally divided by a premaxillary sheet. Th e large dorsal foramen leads directly to th e nasal cavity. Th e second foramen situated ventro-medially leads to the premaxillary sinus. Th e vast nasal cavity has two levels that contact one another only in the posterior part. A wide bony canal located ju st under the skull ro of leads posteriorly from the dorsally position ed external nasal opening. It is called here the do rsal air passage. Th e bony internasal sept um forms its medial wall. The ventral wall of this passage is made anteriorly by the premaxilla. Post eriorly the ventral and lateral walls are formed by ossified, maxilloturbinal and nasoturbinal elements (see below). At the distal terminat ion of the nasal cavity the dorsal air passage bends lat ero-ventrally and reaches the second part of the nasal cavity which is positioned ventro-laterally to the above described dorsal canal. The route of the dorsal air passage is clearl y visible in Euoplocephalus (see COOMBS, MS, fig. 21). The second part of the nasal cavity situated ventrally is broader than the do rsal canal and is located above and posterior to the palatal vacuity. This ventral part of the nasal cavity is interpreted here as a "cavum nasi proprium". As it is located above the palatal vacuity, its anterior part is visible in palatal view of the skull, whereas the posterior part is obscured by the palatine. The cavum nasi proprium is closed anteriorly by a massive protuberance at the level of the lat eral wing of the nasal, maxilla, and lacrimal that form s a base for a bowl-like ossification recognized as a maxilloturbinal (observed also inPinacosaurus). The maxilloturbinal ofSaichania is much stronger in anterior, ventral and median directions. It attains the frontal shelf ofthe max­ illa anteriorly. Medially it contacts the internasal septum. The medial and dorsal surfaces of ex­ pand ed maxilloturbinal form the lateral and ventral wall of the dorsal air passage along the whol e of its length. Posteri or to the maxilloturbinal on the medial part of the cavum nasi proprium there is a labyrinthic ossification, here interpreted as a bony nasoturbinal, which is supported by sagittal crest of the nasal. The crista nasoturbinalis of Saichania is very high and con stitutes the lat ero-posterior wall of the dorsal air passage. Th e horizontal part of the nasoturbinal reaches the internasal septum and form s the posterior ventral wall of the dorsal air passage. A vertical ridge is present lateral to the horizontal part of the nasoturbinal. A deep interconchal recess which extends to the skull roof is located anteriorly between that ridge and the maxilloturbinal. This recess is closed on the lat eral side by a thick scroll-shaped bone which forms a postero-lateral prolongation of the vertical nasoturbinal ridge. This bony element reache s the lat eral wall of the nasal cavity and is located slightly above the postero-Iateral region of the palatal vacuity. As the posterior part of the cavum nasi proprium could not be prepared through the palat al vacuity, the ventral part of the orbito-nasal wall was removed ju st above the vertical wall that closes the palatal vault posteriorly. This revealed a vertical bony crest and thin scroll-like bony sheets in the postero-dorsal part of the cavum nasi. These element s are interpreted here as the ethmoturbinals. I could not trace in detail all connections of the sinus maxillaris with the nasal cavity without partial destruction of th e single skull of Saichania at my disposal. A ventro-medially situated nasal openings leads to a premaxillary sinus that has a connection with the maxillary sinus (see p. 113). In this way a second air passage is being formed that is located below and laterally to the dorsal one. A connection of the anterior part of the sinus maxillar is with the cavum nasi 122 TERESA MARYANSKA proprium has been described. It seems probable that the posterior part of the sinus maxillaris communicated with the nasal cavity as well, and the lacrimal duct opens into the nasal cavity in this region. Another organ, as far unknown in ankylosaurs, was probably present in Pinacosaurus and Saichania. External nasal openings (see p. 103) are horizontally divided by premaxillae. The dorsal foramen is a true air passage. A ventro-medially situated foramen in Saichania and Pinacosaurus leads to a multichambered premaxillary sinus. A bony cavity distinguishable within the premaxillary sinus is situated dorso-medial to other premaxillary chambers and reaches the internasal septum. A deep vascular groove internally on the median wall of this cavity leads to a foramen located on the palatal side of the premaxilla. Similar foramina situated in that region are also present in other ankylosaurs (e.g. Euoplocephalus AMNH 5405, Pano­ plosaurus AMNH 5665, NMNH 11868, NMC 8531). Their existence was reported by GILMORE (1930) but was not explained. The position of this bony cavity in front of the nasal capsule and its connection with the oral cavity show in my opinion that this was a vomero-nasal organ (Jacobson's organ).

Remarks on the nares and the nasal cavities Pinacosaurus grangeri and Saiehania ehulsanensis differ from all other known dinosaurs, and indeed all known fossil and living reptiles, in the unusual structure of their nares and nasal cavities which contain numerous ossifications. However, PARSONS (1970, p. 152) stated: "the nasal anatomy of the various extinct groups of reptiles is almost totally unknown". Distinct bipartition of the paired external nasal openings in Saichania (tripartition in Pinacosaurusi, as well as the complicated structure of the premaxilla within the nares, has no counterpart in other Ornithischia. In gross anatomy the closest structural resemblance of the external nasal region is to be encountered in some mammal-like reptiles, as e.g. some theriodonts (TATARINOV, 1974; BRINK, 1960a, b) and pelycosaurus, in particular Ennatosaurus teeten (TATARINOY & EREMINA 1975).In addition to the premaxilla, the septomaxilla also forms part ofthe narial region oftheromorphs. No separate septomaxilla can be seen in Pinaeosaurus or Saichania, but TATARINOV & EREMINA (1975) interpreted the theromorph Ennatosaurus as having a sep­ tomaxilla contributing to the wall of a cavity for a lateral nasal gland within the lower margin of the nares, an interpretation which could be applicable to ankylosaurs as well. Taking such an interpretation into account, the third, laterally situated foramen in the external narial region of Pinacosaurus (see p. 103) and the smooth surface of premaxilla situated within the external narial region lateral to the air foramen in Saichania may have contained a lateral nasal gland. The exit of the naso-lacrimal duct in that region ofthe nose in Pinaeosaurus and Saichaniawas not observed. Only the entrance of that duct, situated within the orbital wing of the lacrimal, is discernible. It is possible that the naso-lacrimal duct attains the nasal cavity in its posterior part, in the region of posterior part of maxillary sinus. The terminology used herein for describing the structures of the nasal cavities in Pinaco­ saurus and Saiehania is that of mammals and suggests homologies with elements of mammals. Use of mammalian terminology seems justified because the high degree of organisation of the particular elements is comparable to structures of the nasal cavity of mammals rather than to those of reptiles. My opinion is based mainly on the following features: 1) The first bony element to appear (in young Pinacosaurus) is a protuberance of the maxilla, nasal, and lacrimal on which an additional ossification is based. In mammals the maxilloturbinal is an additional ossification based upon the crista maxilloturbinalis, which is the earliest to appear in ontogeny; 2) Bony crests on the internal surface of the nasals in ankylosaurids and therapsida are regarded as a counterpart of the crista nasoturbinalis of mammals and prove the existence of a naso­ turbinal. In Pinaeosaurus as in theriodonts there is only a crest on the nasals corresponding to crista nasoturbinalis, but in Saichania the whole concha nasalis has been ossified: 3) Both ANKYLOSAURIDAE FROM MONGOLIA 123 the location and structure of the third turbinal element correspond to the ethmoturbinals of mammals. The structure of the bony nasal septum is still another argument. In Pinacosaurus the internal bony septum is incomplete and consists entirely of the nasal and premaxilla. In Saichania this septum is fully ossified down the whole length of the nasal cavity. In the degree of ossification of the internasal septum Saichania resembles mammals rather than reptiles. Also, along the nasal suture ofsomemammals (e. g. : Felicidae) the nasal forms an internal protuberance which forms the base for the internasal septum. The proximity of the nasal cavity to the anterior part of the brain resembles mammals more than reptiles (PARSONS, 1956a, 1959b, 1967, 1970). Three conchal formations (preconcha, concha, and postconcha) develop in the cavum nasi proprium only in crocodiles among recent reptiles. The first element that appears in ontogeny is the concha. It is a projection on the lateral caval wall. Later it becomes subdivided into preconcha and concha. The concha is homologized with the maxilloturbinal of mammals (PARSONS, 1970). It seems that the only element of Saichania which is a counterpart of the lateral projection of nasal wall is the maxillo-turbinal. As that element appears in ankylosaurs very early in an ossified state, it may be presumed that it had been present in a cartilaginous state. It is probable that the maxilloturbinal of Saichania corresponds to the preconcha and concha of crocodiles, The postconcha of crocodiles is a convexity of the postero-lateral caval wall. In Saichania the ethmoturbinal system seems to be connected with the dorso-posterior wall. Both its structure and connection with the posterior wall suggest that is has no counterpart in modern reptiles. Homologies of various conchae among the different groups of living tetrapods are uncertain (PARSONS, 1970) and the grounds for homologizing conchal structures of extinct and modern tetrapods are even more ambiguous. The only element of the nasal cavity of ankylosaurids resembling that of some recent reptiles is the dorsal air passage leading from external nares to the posterior part of the cavum nasi proprium. The dorso-medial position ofthat passage, in relation to the cavum nasi proprium with its posterior outlet to the cavum nasi, resembles the location of the vestibulum of some iguanids and agamids, that live in sandy areas (e.g. Dipsosaurus, Callisaurus, Phrynosoma, Uma) (STEBBINS, 1948; PARSONS, 1970). The vestibulum of these genera may form a long and curved tube. This comparison is based only on gross anatomy. It is possible that a large part of the supposed vestibulum of ankylosaurs may have been lined with olfactory epithelium as in the case in some agamids (e.g. Otoeryptis), in which the vestibulum distinguished on the basis ofgross anatomy does not correspond to that distinguished on histological grounds. A different interpretation of the dorsal air passage in Saichania is accepted in this paper. Because of its position between the skull roof and the nasoturbinal, the dorsal passage can be compared with the meatus nasi superior (s. olfactorius) of mammals. This interpretation is supported by the fact that the passage introduced air directly to the olfactory part of the cavum nasi proprium. In this case the ventrally located air passage leading from the additional nasal opening through the sinus maxillaris to the cavum nasi would be a counterpart ofthe meatus nasi medium which brings air to the anterior respiratory part of the cavum nasi proprium. Such an inter­ pretation seems to be supported by a connection of that air passage with the sinuses surrounding the nasal cavity. It follows that the anterior, short and broad part of the nasal duct situated external to the horizontal premaxillary septum which divides the nostrils could be regarded as the vestibulum. The position of the lateral nasal gland suggested above seems to support this view. It may be supposed that the structure ofthe nasal cavity in Pinacosaurus and Saichania des­ cribed above is more or less characteristic for all ankylosaurs. COOMBS (MS, fig. 21), in describing and illustrating sections of the cranial sinuses of Euoplocephalus tutus, has shown paired pre­ maxillary sinuses, nasal sinuses, maxillary sinuses and two pairs of major cranial sinuses. The paired premaxillary sinus, nasal and maxillary sinuses are present also in Saichania. From the position of the transverse section, as shown by COOMBS (l.c.), the major cranial sinuses re­ cognized by him may correspond among others to cavities inside the turbinal ossifications. A dorsalIy located major cranial sinus ofEuoplocephaluscorresponds probably to the interconchal 124 TERESA MA RYANSKA recess of Saicliania, and the ventrally situated one is a section through a lat eral part of naso­ turbinal. The differences are du e to more extensive developm ent of seconda ry bony palate in Euoplocephalus. The structure of the nasal cavity distinguishes ankylosaurids from all other dinosaurs as well as from all other fossil reptil es. Am ong extinct reptiles the Th erapsida have been regarded as closest to mamm als in development of the olfactory apparatus. Since the nasoturbinals in thcromorphs are kn own mainl y in carnivorous forms, the ankylosaurs are the only group of fossil herbivorous tetrap ods (see also p. 139) showi ng a high orga nization of the nasal cavity. Presence of nasoturbinals in theriodonts has been interpreted as evidence of expa nsion of the olfactory surface (e.g. TATARINOV, 1974). In mammals both maxilloturbinal and nasoturbinal play a respiratory role and are not lined with sensory epithelium. As ankylosa ur ids have ossified ethmoturbinals, which undoubtedly executed sensory functi on, it may be supposed that maxilloturbinal and the nasoturbinal in Saichania played only a respirat ory role as is the case in living reptil es and mammals. Similarly, the dorsal and ventral air passages in an­ kylosaurids pr epared the air to enter the lungs as in present-d ay mamm als. I suppose that the dorsal air passage of ankylosaurids was analogous to th e nari al passage in had ro saurs. It is highly probable that within th e had rosaurs cavum nasi pr oprium (e. g. Corythosaurus and Pro cheneosaurus) a system of turbinals was developed, but was not ossified as in a nkylosaurs. I regard int erpretati ons by STERNBERG (1942, 1953) and RUSSELL (1946) of the hadrosaurian crest as a trapping mechani sm more probable than that of OSTROM (1961, 1962) who assumed an olfactory function for that passage. It seems probable that only the posterior part of this crest may be lined with the olfactory epithelium (nea r the olfacto ry part of the cavum nasi proprium), but as in th e ankylosaurs all anterior part pr epared inhaled air and played the respirat ory role. I think HEATON'S (1972) interp retation of hadrosaurian crest as "resonating devices producing individually recognizable cells" (I. c., p. 203) is probabl e, but only as additional and secondary fun ction of that passage.

OSTEOLOGY OF THE POSTCRANIAL SKELETON

Vertebral column and ribs

Cervical vertebrae. Th e entire cervical region (7 vertebrae) is pr eserved in Sa ichania (GI SPS 100/151). Th e atlas and axis are fused with full coossi fication of the zygapophyses and in­ complete fusion of the intercentrum of the atlas and the od ontoid process of the axis (pI. 32, fig. I). The neural arch and th e single head ed atlantal ribs are compl etely fused with a short intercentrum. Th e almost horiz ontal neural arch of the atlas is directed strongly posteriorly. Th ere is no neur al spine. The postzygap ophyses are slightly oblique ant erio rly so that th eir artic ular surfaces ar e almost horizontal. A strong, sharp midventral spine on th e intercentrum is directed anteriorly and ventrally. The body of the axis is stro ngly depressed laterally with a well marked ventral crest. Th e posterior articular face is concave, circul ar and vertical. The neural arch is massive, rising obliqu ely backward with a str ong antero -median crest. The neural spine is flattened antero-posteriorly, laterally terminat ed by swollenings situated antero-dorsal to the po stzygap ophyses. The heavy neural arch is positioned perpendi cular to the posterior part of th e body . Th e neural can al is large. Th e incision s between the po stzygapophyses are bro ad. Th e diap ophyses are sho rt and distinctly ventro-anteriorly directed. The strong axial ribs ar e two head ed and articulate with the diap ophyses and parap ophyses. Cervical vert ebrae from the third to the seventh differ little in genera l structure (pI. 32, fig. 2). Their centra are strongly biconc ave an d depr essed laterally below the diapophyses. Th e broad ventral crest is well marked. Centra of all vertebrae have offset, oblique articular surfaces such that the anterior surface is locat ed below the posterior. Thi s displacement of articular surfaces ANKYLOSAURIDAE FROM MONGOLIA 125 is best developed on the third and fourth vertebrae. The sixth vertebra is almost straight. Gradual changes in the shape of articular surfaces can be noted backward from a circular posterior surface on the axis to a strongly dorso-ventrally flattened and oval one of the sixth vertebra. Neural arches are massive; boundaries of the arches and centra are hardly discernible. The neural canal is high and broad. At 2/3 of the arch length, at the front of each vertebra, the canals are closed by thin horizontal sheets on which a median suture is clearly visible. Massive, transversely flattened, moderately tall neural spines rise upward and obliquely anteriorly at the posterior third of the arches. The neural spines are terminated with tubercular swellings that protrude slightly above the level of the postzygapophyses. These tubercles are separated from the neural spine tuber by a distinct, strongly marked depression on all cervical vertebrae. Pre- and postzyagophyses are prominent; their articular surfaces slant obliquely medially and protrude strongly backward and forward, so that large spaces remain between vertebrae. Diapophyses of all vertebrae are strongly flattened dorso-ventrally and directed laterally and slightly ventrally. Cervical ribs are preserved on all vertebrae. The ribs of the atlas are directed ventro-posteriorly, the axial ribs ventrally and slightly backward and the ribs of third to seventh vertebra are directed latero-posteriorly. Only the broadened distal terminations of the latter are bent forward. Ribs of third and fourth vertebrae have an articular connection. Ribs of fifth to seventh are completely fused . A complete cervical region is not preserved in Pinacosaurus. There are some differences between the cervical vertebrae of Pinacosaurus (ZPAL MgD-II/l) and Saichania: I) in Pinacosaurus the vertebral centra are more oblique than in Saichania and their anterior surface is situated lower than the posterior one; 2) The neural spines on first vertebra of Pinacosaurus are very low and completely reduced on posterior vertebrae; 3) The zygapophyses of Pinacosaurus are longer than in Saichania. The preserved cervicals of Talarurus disparoserratus have massive neural spines and weak diapophyses. Also the ventral crests of the centrum and the supra­ articular tubercles near the postzygapophyses are weakly developed. Different displacement of the articular surfaces of the centra in Talarurus is observed. The anterior surfaces of cervicals are situated higher than the posterior ones, as is in nodosaurids (GILMORE, 1930). Among known ankylosaurids, Saichania is the only known in this moment genus in which the atlas and axis are fused, and Pinacosaurus the only genus in which the neural spines completely disappeared in the posterior cervicals. Dorsal vertebrae. In no Mongolian ankylosaurid is the complete dorsal region perfectly preserved. In Saichania ten vertebrae are preserved. The first dorsal vertebra has a typical dorsal centrum, but its diapophyses and short ribs are coossified with the coracoid. The neural process of that vertebra is located transversely and thus resembles neural spines of cervical vertebrae. The second dorsal vertebra has a short rib backwardly pointing rib fused to the diapophyses. All dorsal centra are long, spool-like, and almost amphiplatyan. Neural arches are well de­ veloped; neural canals high. Transverse processes point obliquely upwards. Prezygapophyses are fused to form single, furrow-like articular surfaces. The postzygapophyses are fused, forming rod-like structure. Starting with third vertebra the ribs are loosely joined. Differences in the structure of dorsal vertebrae of Mongolian ankylosaurids pertain chiefly to the height of neural arches and neural spines. In Pinacosaurus the neural arches are very high as compared to the length of centrum. The neural processes of Talarurus are massive and broadened at top. The degree of devel opment of the ventral crest on the centrum varies. In Pinacosaurus the crest is weakly developed; in Talarurus it is broad; in Saichania it is particular­ ly strongly marked in the anterior part ofcentrum. The posterior dorsal vertebrae are preserved in Pinacosaurus and Talarurus. The centra of four or five last vertebrae are fused, forming the presacral rod. Dorsal ribs. The first ten dorsal ribs are preserved in Saichania (GI SPS 100/151). The first two pairs differs in shape from others and resemble the cervical ribs. The first rib is fused to the coracoid. The short second dorsal rib is directed strongly posteriorly (pI. 34). The third 126 TER ESA MAR YANSKA and fourth ribs are flatt ened antero-posteriorly in ventral part and dorso-ventrally in dorsal. They are directed strongly posteriorly and almost parallel to the vertebral column. Starting with the fifth, ribs are flatt ened antero-posteriorly over their whole length and have a strong convex arch. Ventrally the ribs bend strongly inward so that they embrace the trunk of the animal from the sides and almost meet ventrally. In the do rsal part they are directed laterally. The ribs from the second to fifth are movable, from the sixth rib they are ankylosed to their vertebrae. Starting with the fifth rib a plate-like bony process appears on the posterior margin of the latero-ventral part of each rib. These plate-like processes (pI. 33, figs 3, 4 and 5) of derm al origin overlap the lateral side of the next rib. Each of these plates is fused to the rib at a few points (2-4). The overlapp ing plates form a uniform bony cover over the trunk along a latero-ventral line. Posterior to the fifth rib the position of these pro cesses changes from lateral to ventral. Presumably these processes may occur in other ankylosaurs as well and are not a distinctive feature of Saichania. Similar structures are fragmentarily preserved in Edmontonia rugosidens (GILMORE, 1930) and in Ankylosaurus (BROWN, 1908, fig. 14). Sacrum and sacral ribs. Th e sacral vert ebrae are preserved in Pinacosaurus and Talarurus. Th e sacrum consits of four vertebrae with low, br oad centra which are fused together. The neural canal is broad and high. Th e neural spines relatively low, laterally flatt ened and fused together. In Pinacosaurus (ZPAL MgD-Ij3 1) broad transverse processes of the two first vertebrae are vertical and anteriorly oblique. Processes of the next two are horizontal. Ribs are coossified to transverse processes and contact the ilium along a distance corresponding to the length of the acetabulum. The difference between the sacru m in Pinacosaurus and Talarurus lies in the longer transverse processes and sacral ribs of Pinacosaurus. In connection with this, iliac blades of Talarurus are relatively closer together than in Pinacosaurus in which they are widely spaced.

Scm

Fig. 10 "Dyoplosaurus" giganteus M ALEEV (PIN 551-29). Chevro n of the midd le cauda l vertebra in At -lateral view, A.­ ventral view. The articul ar surfaces shaded.

Caudal vertebrae and ribs. An entire caudal series of vertebrae is preserved in Pinacosaurus (ZPAL MgD-lIj9, ZPAL MgD-lIj31, PIN 614), and a fragmentarytail is kno wn for Talarurusplicatospineus and .Dyoplosaurus". The compl ete caudal region of Pinacosaurus has 40 vertebrae. Centra are slightly biconcave. Proximal vertebrae are of almost the same height and width, and short in comparison with dors al centra. Distally the centra become elongate and lower, their shape ANKYLOSAURIDAE FROM MONGOLIA 127 changes from oval in cross section to square. Natural arches of proximal caudals are very high. Coossification of the arch with the body varies. Neural arches are free in many cases. Distally they become considerably lower and starting with the fifteenth are very low and lack rising neural spines. Dorsal surfaces of the arches become concave. Prezygapophyses are relatively long on the first vertebra and laterally embrace short postzygapophyses. From the fifteenth vertebra the zygapophyses become distinctly V-shaped and their anterior bifurcated part closely embraces wedge-shaped postzygapophyses (pl. 25, fig. 5) along their whole length. The first two or three caudal vertebrae have long, strong transverse processes that contact the posterior part of the ilium. Posteriorly, these processes diminish and by the eight caudal, disappear. Chevrons appear on the third caudal. Chevron shape varies greatly onto the centrum and hang down as V-shaped, laterally flattened trabecules. Distal chevrons become lower, strongly elongated antero­ posteriorly, and flatenned on ventral side. On the fifteenth vertebra at the base of the chevrons bony projections appear which are directed forward and backward to form articulations between the hemal arches similar to zygapophyseal articulations. In this region the chevrons resemble in morphology the neural arches. Ossified tendons closely winding round the caudals appear at fifteenth caudal. Because of the modification of the neural arches and the chevrons, distal parts of the tail have very limited possibilities of movement and form a tail-club, a structure typical in ankylosaurs (text-fig. 10 and 11, Bs). The tail structure in all Mongolian ankylosaurids is similar to that of Pinacosaurus. Dif­ ferences in Pinacosaurus, Talarurus and "Dyoplosaurus" in coossification of some parts of individual vertebrae as well as the degree of fusion of the arches with the centra seem to be individual variations rather than taxonomic characters. In all Mongolian ankylosaurids the tail-club is terminated by bony plates. These are best developed in "Dyoplasaurus" (MA­ RYANSKA, 1970). It follows from a comparison of a young Pinacosaurus with an adult one, that the degree of ossification of tendons and dermal armour plates around the tail may serve as an indicator of individual age. Sternum. The sternal plates of Pinacosaurus and Talarurus were described by MALEEV (1954, 1956). In both, they form a single rhomboidal plate formed by fusion of right and left segments, with a blunt process directed postero-Iaterally which contacts the coracoid. In Saichania (ZPAL MgD-I/112) the sternal plates are more extensively developed (pl. 32, fig. 3 and 4, pI. 33, fig. I). There is the single medio-anterior part which corresponds to the preserved parts of the sternum in Pinacosaurus and Talarurus. Anterior to this part, divergent, relatively short pointed plates are developed, and posteriorly there are strongly expanded plates not completely fused along their midventral line. At least four or five pairs of ribs are connected with the sternum. This very strong ossification of the sternal complex in Saichania is unique not only within the ankylosaurids but also among Ornithischia.

Pectoral girdle and fore limb

Pectoral girdle. The scapula of Mongolian ankylosaurids is completely fused to the coracoid as is the case in North American forms. In Saichania (GI SPS 100/151) the scapular blade (pI. 35, fig. 3) is slightly convex outwards. The broadened distal end of the scapula is bent ventrally. Along approximately half the length of the scapula its dorso-anterior border is flat and broad and at the level of the glenoid forms an acromial process oriented at right angle to the scapular blade. This process is most expanded at the line ofthe coracoid-scapula contact. The ventral margin of the scnapla is broad proximally and sharp posteriorly. The glenoid cavity is formed of the coracoid and scapula. It is shallow, open, and crescent-shaped. The scapula is very low behind the glenoid. The coracoid in Saichania is very small and strongly bent medially behind the glenoid. In natural position the coracoids almost contact one another. A characteristic feature of the 128 TERESA MARYANSKA coracoid of Saichania is its fusion wit h tranverse process and rib of th e first dorsal vertebra. The scapulocora coid of Pin acosaurus (PIN 614) is more convex exte rnally, and th e coracoid is not directed as strongly medially, as in Saichania (pI. 26, fig. 5). The scapulocoracoi d of a you ng individu al (pI. 23, figs 4 an d 5) is relat ively long. That of Talarurus differs mu ch ; the scapula is relatively lon g with a weakly developed acromion p rocess. The coracoid in Talarurus is much larger an d straight and the glenoid is deep an d more closed than in Saichania and Pinacosaurus. Humerus. The hu merus of Saichania is very massive. Its proxima l terminat ion is strongly expanded and almost straight. Med ial to the humeral head a strong process is develcped, and lat erally is th e proximal part of deltop ectoral cre st. T he hu meral head is flattened , lies on the extenso r surface, and wraps aro und onto th e medial p rocess. On th e proximal edge of th e deltopectoral process, ju st lat eral to hu me ra l head , a smooth depression is developed. It re­ sembles a mammalian sulcus int ertubercularis which is co nne cted with biceps mu scle. A stro ng delt opectoral process begins lat eral to hum eral head and exte nds down th e shaft to terminate di stal to humeral mid-l ength (pI. 35, fig. 1). The shaft is po orly separated from deltopectoral cre st, and strongly flatt ened dorso-ventrally. The whole bon e is slightly conv ex dorsally. The distal termination is less expanded th an th e pr oxim al. The distal articular surface lies on flexor surface. Both epicondyles are developed. The radial con dy le is ova l and situated at the centre of distal expan sion. The uln ar groove is relatively shallow. A broad articular surface for th e uln a is also develop ed on th e flexor surface of th e entepicondyle. This surface wraps aro und onto th e flexor sur face adjacent to th e cap itul um . The supinator process is prominent. Axes of th e proximal and distal expansions lie at an angle of 15° to each other , so torsion along the shaft is present.

The hume rus of Pinacosaurus (text-fig. 11 , B 1 pI. 23, fig. 1; pI. 26, fig. 4) is more slender th an in Sa ichania and Talarurus as a result of a lesser expansio n of the proximal termination and, especially, of the med ian process. Also, th e deltopectoral crest is less developed, especially in its proxim al part. The humeral shaft is bett er distinguished and less flatt en ed tha n in Saichania. The ulnar groove in Pinacosaurus is bett er marked. The humeral head is developed on extensor surface but wraps aro und onto th e proximal end. The humerus of Talarurus is more massive than th at of Pinacosaurus, but not so massive as in Sa ichania, with a less develop ed proximal expan sion and a very well developed ulnar groove. The humerus of Saichania is more massive th an in the N orth American Euoplocephalus a nd An kylosaurus, and has a mor e expanded proximal end and a larger delto-p ectoral crest. A strongly concave, smo oth surface on the proximal margin, lat eral to humeral head is ch aracteristic of Saichania, and is not present in other ankylosauri ds. In Saichania it is developed in a place which in Euoplocephalus and An­ kylosaurus, acco rding to COOMB S (MS) is the insertion of scapular delt oid mascle. Both the shap e of th e humeral head an d th e presence of a bon y connection between the pectoral girdle and the vertebral column prove th at th e m otion mechanics of th e fore limb of Saichania must have been different fr om th at of other ankylosaurs.This probl em will be dealt with in another paper. Ulna. The ulna of Sa ichania is very massive with a broad proxim al termination an d an antero­ posteriorly flattened distal end.The olecrano n process is re latively low. The sigmoidal notch si strongly ope n and sha llow.The processes bordering it meet at an angle of about 120°. The posterior process is more extensive th an th e lateral,The proximal articu lar surface is very large, and orie nte d at an angle of 30--40° to the longitudinal axis. Becau se of a very strong develop­ ment of th e posterior process of the proximal part of the uln a, th e oblique ly situated articular surface embraces about 2/3 of the ulna r length (pI. 29, fig. 5). The distal terminatio n is narrow, the distal margin sharp, an d th e articular surface is oblique latero-medially and parallel to the proximal articulatio n. The ulna of Pinacosaurus (pI. 23, fig. 2 ; pI. 26, fig. 3) differs fr om tha t of Saichania in th e lesser exp ansion of its proximal part a nd in having a distinct shaft which embraces about 1/2 of the ANKYLOSAURIDAE FROM MONGOLIA 129

Fig. II Diagrammatic drawing showing the outlines of some bones in Asian Ank ylosauridae. A - Talarurus plicatospineus, B - Pinacosaurus grangeri, C- Saichania chulsanensis, I - left humerus in dorsal view, 2 - right ulna in posterior view, 3.- proximal surfaces of the left rad ius, 4 - proximal surfaces of the left meta carpals, 5 - the chevrons of caudals (third to sixteenth from left to right) . Not to scale.

ulnar length. The olecranon is very tall and sharp. The sigmoidal notch is less open than in Saichania and the processes bordering it are equally developed and meet at an angle of 70°. The olecranon is less developed in a young Pinacosaurus (ZPAL MgD-Ilfl) than in an adult (PIN 614). The ulna of Talarurus is more massive than in Pinacosaurus but more slender than in Saichania, except that its distal terminati on is more expanded and articul ar surface more terminally positioned (pI. 26, fig. 2; text-fig. 11 , 2). Radius. The radius of Saichania is straight, massive, slightly broadened at proximal ter­ mination, and strongly expanded distally. The proximal articular surface is oval, slightly concave 9 - Palaeontologla P olonlca No. 37 130 TERESA MARYANSKA and terminally positioned. The bone is flattened antero-posteriorly, The entire dist al termination of the radius protrudes considerably beyond the ulna. Thi s protrusion together with the oblique distal articularsurface ofthe ulna ind icates a strong oblique orientation in the forearm. The radius of Pinacosaurus (pI. 23, fig. 3) is more slender th an in Saichania. The proximal articular surface is very narrow (text-fig . 11, 3) especially in the young individual. The radius of Ta/arurus is almost as massive as in Saichania. Manus. Aside of the intermedium mentioned by MALEEV (1954a) none of the carpal elements are present in any Mongolian specimen, Metacarpals I-V are pre served in natural arrangement in Saichania (GI SPS 100/151) and in Pinacosaurus (ZPAL MgD -II j9). The metacarpals I -V of Saichania (pI. 36, fig. 2) are very massive and flattened antero-p osteri orly. Metacarpal I is the widest; its proximal articular surface (text-fig . 11 ,4) is slightly concave and subquadrangular. A strong protuberance exists on the posterior side of the pr oximal end of the shaft. The distal articular surface has a strong iternal condyle. Metacarpal II ha s a subtriangular, slightly concave proximal articular surface. A broad crest extends upward and outward from the shaft on the side of internal condyle. The distal articular surface has equa lly devel oped condyles. Metacarpal III is the longest. Its proximal surface is strongly convex and oblique posteriorly. The shaft is flattened with a broad crest running on the posterior side from internal condyle. A distinct tuber is present on the slightly concave anterior side of the shaft near the lateral margin. Distal surfaces of metacarpals I and II are convex with well developed condyles and are situated in a plane perpendicular to the long axis of the bon e; the articular surface of metacarpal III is distinctly shifted to the anterior side of bone. Metacarpal IV is flattened laterally and has crescent-shap ed, convex proximal surface. A strong depression in the shaft is present proximally on the lateral side cf metacarpal V. Metacarpal V is the smallest of all metacarpals. Th e distal termination is narrow, the condyles not developed. In natural arra ngement the metacarpals form an arch with metacarpal V shifted distir.ctly backward. Not all the phalanges are preserved. Th ey are very sho rt with shallow pr oximal articular surfaces. Their distal terminations have strongly divided condyles. L'gamental depressions are not developed. The distal articular surface is more expa nded on the dorsal than on the ventral side, suggesting greater extension than flexion movements. At metacarpal I a partial and at metacarpal II a complete small crescent-shaped bone is preserved, each having a concave articular surface for the metacarpal and convex dist al articulation. These bones articulate with the metacarpals to form an extension of distal articular surface on the anterior side. I re­ cognize them as sesamoid bones (pI. 36, fig. 4). Th ere is no trace of sesamoids at metacarpals III and IV in which the distal articular surfaces are well developed anteriorly. The general shape and po sition of articular surfaces of metacarpals in Pinacosaurus (ZPAL MgD-IIj9) is similar to that of Saichania. Slend erness of Pinacosaurus metacarpals (pI. 24, fig. 5) is an essential difference. First phalanges are pr eserved at metacarpals I -IV; they are long, ulike those of Saichania, and have strongly developed lat eral condyles. The second phalanx of digit III is low and the ungual ha s the form of a very small nail. Th e ph ala ngeal formula ofPinacosaurusgrangeri is unknown. It can be only stated that its third digit has three phalanges, not four as it was claim ed by MALEEv (1954). Most pr obably a phalangeal formula of that genus was 2 : 3 : 3 : 3 : 2 as in Talarurus. Ccmparison of Saichania, Pinacosaurus and Talarurus shows that the manus cf Saichania was most massive, that of Ta/arurus more slender, and the longest and most slender was that of Pinacosaurus.

Pelvic girdle and hind limb

The pelvic girdle and hind limb in Saichania (GJ SPS 100/151) are not preserved. D escriptions of these in Pinacosaurus and Talarurus by MALEEV (1954, 1956) are supplemented below. In Pinacosaurus and Ta/arurus the subacetabular and postacetabular portions of the ilium lie horizontally, whereas the narrow preacetabular part of the ilium is oblique, particularly in ANKYLOSAURIDAE FROM MONGOLIA 131

Pinacosaurus. A thin, narrow, dorsal plate is situated hori zontally , and th e massive preacetabular ilium is locat ed along d orso-Iat eral bod y line. Strong divergenc e of th e prcacetabular ilium is characteristic of Pinacosaurus, and suggests th e tru nk of that genus was very wide. The aceta­ bulum ha s a more prominent border in Ta/arurus, whereas in Pinacosaurus it is hardly marked. Bony connecti on of the ilium wit h sacral vertebrae in both genera tak es place along the length of th e acetabulum. The first sacral rib joins the ilium at th e front of the acetabulum, th e fourth ­ behi nd its posterior border. The ischium of Ta/arurus differs from tha t of Pinacosaurus in being wide r pr oximally and having a greater backward curvature of the sha ft. The mat erial under study does not furnish any new data about the structure of th e ilium and hind limb in comparison with th e descr ipt ion of MALEEV (1954, 1956). It mu st be corrected that the femur illustrat ed by MALEEV (I954, p. 158, fig. I I) do es not belong to Syrmosaurus (= Pinacosaurusi but to Talarurus (PIN 517-101). P es. Fragments of th e pes are preserved in 'the ske leto n of Pinacosaurus (ZPAL MgD-II/9). Metatarsal I is preserv ed together with the ph alanges of digit I as well as the fragments of metatarsals II and III and ph alanges of th e fourth digit. The ph alangeal formula is unknown, but it is po ssible to stat e th at th e pes of Pinacosaurus has four well developed metatarsals (pI. 24, fig. 5), not three as stated by MALEEV (I 954). The preserved phalanges are very low with strongly developed articular sur faces th at are deeply concave proximally an d convex with double condyles distally. Ligam ental fossae are distinctly marked. The terminal hoof-like unguals (pI. 24, fig. 7) are large and broad with well develcped proximal articulations. As compared to th e pes of Ta/arurus (MALEEV, 1956), metatarsals of Pinacosaurus are more slender and the unguals br oad er. A ph alangeal formula of 2 : 3 : 4 : 5 : 0 for Ta/arurus was proposed by MALEEV (I.e.) , but I doubt th is is correct. The pes which served as a basis for this formula was completed with bones belonging, in fact, to severa l individuals, and it is likely that the fourth digit consisted of four, not five, phalanges. The complete pes of " Dyop/osaurus" is unknown. The preserved fragments suggest th at it has four digits as in other Mongolian ankylosaurids.

Dermal armour

A complete armour of the anterior part of th e body in natural arrangeme nt is preserved in Saichania (GJ SPS 100/ 151). It consists of two un ifo rm belts , a cervical and a pectoral, an d numerou s loose elements that cove red the body both dorsally and ventrally. Cervical and pectoral half-rings have identical construction. They consi st oftwo bony layers (pI. 36, fig. I). The deeper half-ring is a uniform, bony bar thickened at its end. Thi s de eper layer is covered by a superficial half-ring formed of three pairs of large, keeled , sha rp ly po inted plates that are joined one with another by square coossified fields of fine, oval, tubercular ossifications. These two layers of each half ring are fused together along th e anterior margin at 1/3 of its length. The central and distal parts of th e belts are free. Both layers are fused as well by bony bridges along the posterior margin. Keeled plat es th at tak e part in the structure of the external layer of the pectoral girdl e are much larger th an th ose of the cervical girdle and sma ll fields of fine ossifications are not so closely set as in th e cervical half-ring. Posterior to the pectoral girdle th e armo ur plat es are symmetrically arra nged into half-rings and longitudinal belts (see KIELAN-JAWOROWSKA & BARSBOLD, 1972, pI. 2, fig. I). However, they float ed freely in th e skin and did not possess a continuous lower bon y lay er. Many types of plates and scales may be distinguished amo ng th e armo ur elements (pI. 36, figs 6-1 I); the mo st common are : I) large keeled thin-walled, sharp-po inted plat es ; 2) smaller, keeled thick­ walled plates; 3) conical scales slightly hollowed at the base; 4) cre st-like, laterally flattened large scutes of narrow base and sha rp, peripherally situated peak; 5) oval, low, asymmetrical sharp-po inted scales; an d 6) various fine ossifications. Almost everyone of the larger plates 132 TERESA MARYANSKA is surrounded by a ring of fine tubercular ossifications. In to p view the different types of scales are set in symmetri cal longitudinal belts to th e sides of th e median bcd y line. Symm etrical conical scales evidently resting on the neural spines are placed along the median line. Lateral to them is a belt of large, keeled, thin-walled plates with thei r peaks pointing backwards. These large plat es rest on ribs and int ercostal ten dons. Interspaces are covered with numerous fine ossifications. The sides of the body are very strongly armoured by plates and scales arranged in several rows. Each large element is surrounded by smaller, conical scales. Flattened ke eled plat es are arranged midway on th e flanks of th e body with th eir peak s pointing backwards. Ventral to them th ere is a row of large plates, the peaks of which also point backwards, and below thi s there is ano ther row with peak s pointing forward . The armo ur of Pinacosaurus was probabl y similar in str ucture as may be determined from preserved elements. Both two-layered half-rings, cervical and pectoral, are distinct in a young individual. The armour of Talarurus consists of cervical an d pectoral half-rings, but they are narrower th an in Saiclzania, and similar pelvic armo ur. R ib-groove ornam entation is cha­ racteristic in the armo ur elements of Talarurus. Armour plat es of Sauropliles (BOHLIN, 1953) are ornamented in a similar way. The preserved armo ur element s of "Dyoplosaurus" have been described (MARYANSKA, 1970).

GEOGRAPHICAND STRATIGRAPHIC DISTRIBUTION OFASIATIC ANKYLO SA URIA

There are at least 26 Asiatic sites from which the remains of ankylosaurs are kn own, in the sediments of the Lower and Upper Cretaceous. All these sites are situated roughly between 63°-130° east of Greenwich and 23°-62° latitude north, in th e area of the Mongolian People's Republic, China, USSR and (?) Ind ia (text-fig. 1 and table 1). The present kn owledge of continental Cretaceous sediments of Central Asia does not allow precision in determining relative and abso lut e age. Table 1 shows a compilation of recent data. Temporal positions are based on studies of continental faunas (invertebrat es, mammals and som e cases of din osaurs) which in most cases are endemic to Asia, and therefore the strati­ graphic conclusions must be regarded as tentative. Only three Upper Cr etaceous units can be formally accepted. These are : The Djad okhta Format ion (BERK EY & MORRIS, 1927; LEFELD, 1971), Barun G oyot Format ion and N emegt Formation (MARTINSON et al., 1968; GRADZINSKI et al., 1969; GRADZINSKI & JERZYKIEWICZ, 1972). Stratigraphic units called "svita " distinguished by Soviet geologists in Mongolia do not always correspond to units of intern ati onal classification, and consequently an informal term inology is here applied to such units (e. g. Bayn Shireh "for­ mati on" - svita). A compilation of recent results obtained by the Soviet geologists in Mongolia has been present ed by M ARTINSON (1973, 1975) who has distinguished the following svitas in th e Upper Cretaceous: Sayn Shand (Lower Cenomanian), Bayn Shireh (Upp er Cenomanian to Lower Santon ian), Baru n G oyot (Upper Santonian to Campanian) and Nemegt (M aas­ trichti an) . MARTINSON (I .c.) does not distinguish a sepa rate stratigraphic unit for th e sediments of, amo ng others Bayn Dzak (Djad okhta Formation of Am erican and Polish au thors), but has included them into the Baru n G oyot svita.BARSBOLD (1972) in the stratigraphic scheme of the Upper Cretaceous of Mongolia mentions the Djadokhta svita isochronous with th e Barun G oyot svita . According to the results of th e Polish-Mongolian expeditions to the G obi Desert in Mongolia (GRAD ZINSKI et al., 1969 ; LEFELD , 1970 ; GRADZINSKI & JERZYKIEWICZ, 1972; KIELAN-JAWOROWSKA, 1974a, 1975 a, 1975 b; MARYANSKA& OSM6LSKA, 1975) both the litol ogy and faunal contents of Bayn Dzak sediments prove without doub t th at th ey should be regarded as a separate lithostratigraphic and biostrat igraphic unit (Djad okhta Format ion). older than the Barun Goyot Formation. ANKYLOLAURJDAE FROM MONGOLIA 133

Sediments at Bayn Shireh that contained remains of Talarurus plicatospineus were regarded by MALEEV (1952e) as the youngest Cretaceous sediments in Mongolia, i.e. Maastrichtian in age. This opinion was not confirmed by later investigations. An Upper Cenomanian to Lower Santonian age ascribed to these sediments by Martinson (1973, 1975) is more probable. The same pertains to Sheeregeen Gashoon sediments, the age of which was initially determined by MALEEV (l952e) and KONZHUKOVA (1955) as Maastrichtian. According to new data (MAR­ TINSON, 1973; BARSBOLD, 1972; SHUVALOV, 1975) these sediments or at least a part of them, re­ present the Bayn Shireh svita and are Upper Cenomanian to Lower Santonian in age. According to MARYANSKA & OSM6LSKA (1975) the sediments at Sheeregeen Gashoon containing Micro­ ceratops gobiensis (and remains of ankylosaurs) should be regarded as older than the Djad­ okhta Formation which yields Protoceratops andrewsi. The oldest sediments containing ankylosaurs in Asia are at Khovboor (see text-fig. 1 and plate 1) in Mongolia (undescribed remains in the collections of PIN) and at Tebch in China (Sauroplites seutiger, BOHLIN, 1953). At all these sites the ankylosaurs are associated with, among other things, psittacosaurs, which suggests a Lower Cretaceous age. Undoubtedly younger are the sediments of the so-called Bayn Shireh svita containing Talarurus plieatospineus and, among other things, remains of primitive hadrosaurs. The Djadokhta Formation with Pinaeosaurus grangeri, other dinosaurs, and mammals (table 1) seems to be still younger. The Barun Goyot Formation and its chronological counterparts contain Saiehania ehulsanensis gen. n., sp. n. and are younger than the Djadokhta Formation. The youngest Cretaceous se­ diments in Mongolia containing ankylosaurs ("Dyoplosaurus" giganteus) are those of the Ne­ megt Formation. The above scheme of relative geochronology ofCretaceous sediments is based on investigations of ankylosaurs. It confirms the results of KIELAN-JAwOROWSKA (l974b, 1975a, 1975b) based on studies of mammals and the data obtained by MARYANSKA and OSM6L­ SKA (1974, 1975) based on studies of the Pachycephalosauria and the Protoceratopsidae. This scheme agrees only in part with suggestions of MARTINSON (1973, 1975) based on studies of invertebrates. The ankylosaurs are represented in Asia only by the Ankylosauridae (sensu COOMBS, MS). The Nodosauridae (sensu COOMBS, I.e.) are unknown from Asia. Abundance of Ankylosauridae in post-Middle Campanian Late Cretaceous sediments has been noted (e.g. GILMORE, 1923, 1930; NOPcsA, 1928; PARKS, 1924; SAHNI, 1972; STERNBERG, 1929; COOMBS, MS) in North America (Two Medicine Formation, Oldman Fm., Judith River Fm. Edmonton Fm.). In Asia, Ankylosauridae are present earlier in the Late Cretaceous or even in the Early Cretaceous. The earlier appearance of Ankylosauridae in Central Asia than in North America may suggest that the former area is the center of their early radiation. Comparison of the latest Cretaceous ankylosaurian fauna of Asia and North America (Santonian - Late Campanian or Maas­ trichtian in Asia and Middle Campanian - latest Maastrichtian in North America) shows that in this interval of time the Ankylosauria were represented on both continents not only by different genera of Ankylosauridae but partly also by different families. KIELAN-JAWOROWSKA (1974a, 1974b) on the basis ofstudy ofCretaceous mammals from both continents has concluded that Asia and North America during the latest Cretaceous were separated by a barrier, possibly by marine straits, and interchange of mammals had taken place by means of a sweepstakes route. Existing currents allowed occasional migration of mammals from Asia to America but not vice versa. Nevertheless, the problem ofmigration ofdinosaurs is more complicated. The earli­ est record of an ankylosaurid in North America is in the Middle Campanian. Approximately at the same time the Protoceratopsidae and Pachycephalosauridae made their appearance on that continent. Possible migration of these dinosaurs from Asia America may have taken place prior to the Campanian. The study of this problem needs detailed investigations of the relations between the Asiatic and North American dinosaurs such as Saurolophus angustirostris and S. os­ borni and the genera Tarbosaurus and Tyrannosaurus and Albertosaurus. ROZHDESTVENSKY (1968, 1974) suggested a close relationship between these forms and bidirectional migration 134 TER ESA MARYANSKA of dinosaurs between Asia and America during po st-Campanian times. The qu estion of Early Cretaceous migrations of dinosaurs between the continents poses another pr oblem. Absence of certain groups on one or an other con tinent neith er proves nor precludes migration (sensu land conn ection) but rather pr oves existence of other barri ers (e. g. ecological ones) that have hampered or prevented migrations of certain groups. For example, the Nodosauridae are well represent ed in the Cret aceous both in Eur ope (since the Wealden) and in North America (since the Arundel and Cloverly Form ation s). The Nod osauridae pen etrated into Asia neither during Early nor Late Cretaceous, although the Pachyc cphalosauridae and the Iguanodontidae did. But the olde st and most primitive representative s of both latter families occur.in the Wealden of Europe tog ether with the Ncdosauridae, Migration of pachyc ephalos aurs and iguan odonts from Eur ope to Asia proves that such routes were possible. However, they were used neither by the Early Cretaceous Eur cpean ncdosaurids and hypsilcphcdonts, nor by the Asian psittacosaurs. Possibly fauni stic differences between Central Asia and other continents during Cretaceous times depended on different ecological cond itions conn ected with distance from the sea and difference s in vegetation. Possibly the absence of the Ceratopsidae in the late Cretaceous of Asia (taking into account the differentiation of the Protoceratcpsidae - MARYANSKA & OSM6LSKA, 1975) is connected with unfavourable livnig conditions rather than with lack of opportunity for migrati on.

EVOLUTION OF ASIATIC ANKYLOSAURIDAE

A comparison of the relatively rich oste ological material of the Mongolian Ankylosauridae beginning with Talarurus, through Pinacosaurus and Saichania to "Dyoplosaurus" reveals some trends in the evolution of their skeletons (text-fig. 12). The se are: the changes in skull structure toward greater massivene ss attained by shortening relative to width, by stronger ossification of the orbital region involving the neomorphic bones, and by stronger ossification of the palate. The latter is attained through develcpment of horizontal maxillary shelves and expansion of the palatine forward and the maxilla backward. Th e tendencies toward greater massiveness of the skull are reflected in the structure of the neurocranium. Fo rms with relatively light skull (Talarurus) are characterized by separate openings for nerves and blood vessels posterior to the foramen ovale. In intermediate forms (Pinacosaurus) a reducti on in the number of openings is observed and in the most advanced forms (Saichania, Tarchia) a single, common opening is formed. A complication of cranial sinuses parallels these changes. This is expressed by development of multichambered premaxillary and maxillary sinuses. Studies of Mongolian Ankylosauridae do not support the generally accepted opinion on a tendency to general increase in body size (STEEL, 1968). Not a single form in Mongolia has attained size of the North American ankylosaurs such as Ankylosaurus or the large st individuals of Euoplocephalus. This pertains both to skull size and body size. A general tendency toward greater massiveness existed in the postcranial skeleton, but it was not consistent through all form s. For example, Pinacosaurus is characterized by a relatively light skeleton and slender limbs, but Saichania is characterized by a very massive skeleton. The oldest well kn own Upper Cretaceous ankylosaurid (Talarurus) occupies an intermediate position with regard to massiveness of the skeleton. A comparison of the postcranial skeleton shows that no essenti al change occurred in pelvic structure from Talarurus to "Dyoplosaurus". The pectoral girdle has been subjected to much greater changes. The sternum has changed con siderably. A comparison of Asiatic Ankylosauridae with North American genera, generally younger stratigraphically, suggests that form s similar to Talarurus and Pinacosaurus may have migrated from Asia to North America. Th ese genera could not have been more advanced than Pina­ cosaurus. North American ankylosa urids (Euoplocephalus and Ankylosaurus) have more progressive ossification of the palate and further shortening of the skull in relation to its width. ANKYLOSAURIDAE FR OM MONGOLIA 135

ASI A NORTH A MERI CA Formati on Speci es Format io n Species

c-'" '" L.. a Q) ' Q) ~ ~ c-"'-au ~ o E Ankylcsaurus 0> ::::»-0= l1> U' lU 1':1 ~ ; n i ~cn t r i s '" :J: E •Duo plosnu rus u C Q) .E c z giga ntE :.lS a E -0 LU c, Q) (f) \ ~ 0 ....J Euoplcceph nlus ~ u '" a tutus a :n 2 a C (!) ~~ ~ Saichania TJhiU c: l1> '" .- t'l c 2 u ch ulsanen sis kielanae '0 C c "'d '" . - ~ Q) ~ "'C ._- 0 c o; Q) I 0 ° 0" t:l ~ CD '" ::::;; -cc ":i'" I a c 0" I '" ~ Q) L.. \ / l- I U I o (f) ~VJc ...:at:~~ ~: '""'"' a 0 gra ng erl -c -C,. c c: .- ~ .... '-' a - <=l '" -- '" ------"'- ..c. III 7Tal arurus . ~ ~ o, (J) (/)J:;~.- -c :::> -es -", T,'plic''"'"'atospin eus ~ ~ . ~ 'y ""'W o r:r co Q) I I \ '"::J Sauraplites scutiger a and undescribed species Q) u from Khovboor c ~ I l1> L.. , U I L.. l1> I ~ 1 0 ....J I L-- Fig. 12 Distribution of the Ankylosauridae in Asia. But, they show some new tend encies such as a different structure of the nasal region and body size increase. COOMBS (MS) suggested that Euoploeephalus, Talarurus and Asian"Dyoplosaurus" are congeneric. A detailed comparison of these forms shows that they represent separate genera, presumably of common origin. There is no basis for the suggestion that the same ankylosaurid genera are present in Asia and North America as e.g. "intercontinental" distribution of Dyoplo­ saurus (CHARIG, 1973) and Euop locephalus (COOMBS, MS). The present study of the Asiatic Ankylos auridae, similar to that for North America made by COOMBS (I.e .), does not elucidate the relationships between the Nodosauridae and the Ankylosauridae. One of the oldest Mongolian ankylosaurids - Talarurus, is a typical member of the Ankylosauridae. It shows some similarities to the Nodosauridae, such as a relatively long skull, subspherical occipital condyle, incomplete coossification of distal caudal vertebrae. Also, some of these are probably primitive characters, retained in Nodosauridae but lost in Ankylos auridae, for example with development of tail club. All these characters, probably 136 TERESA MARYANSKA primitive in Talarurus, are, however, insufficient to prove an origin ofthe Ankylosauridae directly from the Nodosauridae. New data on the origin of the Ankylos auridae will be supplied by a study of an Early Cretaceous specimen of a primitive ankylosaur from Khovboor.

POSITION OF ANKYLOSAURIA WITHIN ORNITHISCHIA

The origin of the Ankylosauria and their relation to other Ornithischia are stilI an open question; MARSH (1895), LAPPARENT and LAVO CAT (1955), HUEN E(1956), SWINTON (1970) and others, classified ankylosaurs with stegosaurs (Stegosauria, Thyreophora, Orthopoda), whereas ROMER (1927, 1968) put them into a separate suborder, Ankylosauria. The basis for uniting ankylosaurs and stegosaurs, were common characters such as quadrupedality, the presence of armour, and tooth morphology. ROMER (1968), who was skeptical about a possibility of a com­ mon origin of the Ankylosauria and Stegosauria stated (I.c., p. 143): "It is possible that there may have been an early armoured common ancestor of stegosaurs and ankylosaurs, but there is no reason to believe that there was, and no evidence for it". COOMBS (MS) pointed some common features of both groups such as: presence of armour; lack of an obturator process; lack ofa premaxilla-lacrimal contact; tooth morphology; limb proportions; phalangeal formula; structure of the metapodia; as well as the morphology of the palpebral bones. According to COOMBS (MS) some characters that are common to the Ceratopsia and the Pachycephalo­ sauria suggest a derivation of those groups from quadrupedal ancestors and are not a basis to unite the Ankylosauria and Stegosauria into one natural group. The Ankylosauria and Stego­ sauria are the only ornithischian groups in which we cannot show that their quadrupedality is secondary. There is evidence that the Ceratopsia derive from an ornithopod line and are secondarily quadrupedal (ROMER, 1968; MARYANSKA & OSM6LSKA 1975). ROMER'S (1968) conviction about secondary quadrupedality of Stegosauria resulted from the considerable differences in length offore and hind limbs. However, the relative length of hind limbs increased in the group'with time (STEEL, 1969). CHARIG'S (1964, 1972) hypothesis of archosaur locomotion and evolution of archosaur pelvis and hind limbs also supports the theory of primary quadru­ pedality for Stegoauria and Ankylosauria. In light of the data about the Mongolian ankylosaurs, one cannot dismiss the possibility of a common derivation of the Stegosauria and Ankylosauria separate from the ornithopod line. This seems to be supported by the following common characters: lack of a premaxilla-lacrimal contact; primitive tooth structure; convergence in general skull structure; presence of two palpebral bones incorporated in the skull roof; presence of a postfrontal; lack of the obturator process; completely (Ankylosauria) or partly (Stegosauria) closed aceta­ bulum; presence of armour; and a primary quadrupedalism. According to ROMER (1956) the postfrontal does not occur in the Ornithischia. In fact, this bone does not occur in the repre­ sentatives of broadly termed onithopod line, but as evidenced by Stegosaurus (GILMORE, 1914) and Pinacosaurus (MARYANSKA, 1971), one must accept its existence in the Stegosauria and Ankylosauria. According to OSTROM (1970) and COOMBS (MS) the pelvis of all the Ankylosauria has a closed acetabulum, no anterior pubic process (weakly developed in Sauropelta), and the pubis is significantly or completely reduced. The ilium differs from that of other Ornithischia. No pelvis so far known among Triassic and Jurassic Ornithischia is a suitable predecessor to that of the Ankylosauria. The oldest Ornithischia (e.g. Fabrosaurus) have a fenestrated ace­ tabulum. CHARIG (1972) assumes that the acetabulum of ankylosaurs is secondarily closed. However, there is no proof of this in known ankylosaur material. It seems probable that the closed acetabulum is a primary character of ankylosaurs. According to CHARIG (1972) the vertical limb posture in dinosaurs is always accompanied by fenestration of the acetabulum, ANKYLOSAURIDAE FROM MONGOLIA 137 by development ofa very distinct and strong, inwardly turned femoral head, and by development of digitigrady. The limb posture of the ankylosaurs is different from that of quadrupedal cera­ topsian and sauropods. The femoral head of ankylosaurs is less developed than that of other Ornithischia and is only very slightly turned inwards. The fore limbs are flexed, in some ankylo­ saurids very strongly (Saichania), with an almost horizontal and laterally directed humerus. The stance and gait of ankylosaurs did not necessitate fenestration of the acetabulum. Primary primitiveness of some pelvic characters in ankylosaurs may be an additional argument against connections between the Ankylosauria and the ornithopod line of the Ornithischia. If the cha­ racters shared by ankylosaurs and stegosaurs argue for their common origin, the difference in the structure of pelvis must be considered. Both Stegosauria and Ankylosauria are cha­ racterized by greater development of the anterior process ofthe ilium compared to the posterior process. Differences in morphology of the anterior process of the ilium probably originated

:>:. ... L ~ . ~ er-. ::,.' V"l -e -. :::> ~ .g:r:::: :. '"LU . Ul ', ~ ' . U I-- .:ri. ..:L.: , « L ~ ~ . I- ~ :..; ~ ~ : .~: : u , ~ " L ',\ ~ , a- \ \ ~ , ~ , U , \ . ~ , , ~ ...... iii ... , , '\ " , ' ... , ..... , , ..... , ,, '- , ..., " \ \ 1:;:::';':::;:1 Ankylosouria ..... \ \ ...... " \ \ ~ Scelidosouridae " H. \ Stegasauria L " ~ ~ a­ Pachycephalosauria u a. =:J ~ Ceratopsia V"l - V"l ~ Ornithopoda 7 fTTl?TTTl Family incerlae Ul..ll.ll.U subordinis 8-=dThecodontia

Fig. 13 Supposed phylogenetical relationsh ips of the ornithischian dinosaurs. after the definitive separation of the two groups and are correlated with the structure of the pubis. The anterior process of the pubis in stegosaurs might have developed in connection with large body size and a different mode of locomotion than in ankylosaurs. Stegosaurian verti­ cal limb posture resulted in partial perforation of the acetabulum. It follows from studies of the Ceratopsia that the anterior process of the pubis changed relatively quickly in various protoceratopsids and ceratopsids, being extensive in some groups and weakly developed in others. Pelvic changes in stegosaur evolution were directed differently than in ankylosaurs. One may presume that the closed acetabulum of the Ankylosauria is primary and that partial closure of acetabulum in e.g. Omosaurus, Stegosaurus (GILMORE, 1914) or Wuerhosaurus (DONG-ZHI-MING, 1973) is primary as well. It follows that stegosaurs did not reach complete fenestration of the acetabulum. Closure of the acetabulum in some Ceratopsia is similar to that in Stegosauria, and should be considered as a result of secondary adaptation to quadrupedality. It is difficult to decide whether the Ornithi schia are diphyletic or whether their subdivision into two branches took place immediately after their monophyletic origin from a thecodont 138 TERESA MARYANSKA

ancestor. THULBORN who acce pt s mo nophyletic origin of th e Orn ithischia wro te (1971, p. 77): "This hypsilophod ont plexus is fundamental to the whole of ornithischia n ph ylogeny; it represent s th e ancestry, ultimately at least, of such gro ups as the iguancdonts, hadrosaurs, and cerat opsian s". Only the origin of the iguanodonts, hadrosa urs and ceratopsians fr om th e hypsilophod ont line can be proved. Lack of evidence on derivati on of the an kylosaurs an d stego­ saurs from hypsilop hodon tids d oes not p reclude diphyletic origin of the Ornithischia. G ALTON (1974) assumed th at the Upp er Triassic Fabrosaurus sta nds close "to the archetypical or­ nithisch ian fr om which all other ornithischian origi nally derived". If, however, all k nown T riassic Ornithischia (e.g. Pisanosaurus, Fabrosaurus, Heterodontosaurusi are not on ly bipedal but also cursori al (GALTON, 1974), th en derivati on of th e early Ju rassic, quadrupedal S celidosaurus from forms relat ed to Fabrosaurus seems doubtful. Secondary quadrupedality had appeared in Ornithischia not earlie r th an th e Lat e Cretaceou s. Ancestors of the an kylosauri d line are expected amo ng quadrupedal Pseud osuchi a which might have produced two lines of ornithischians. Acc ept ance of primary bipcdality for all p rimitive Ornith ischia (GALTON, 1973) also seems imp ossible becau se of reasons similar to th ose given above. Suggestions of WALK ER (1961) abo ut similarities between Aetosauria and Ornithi schi a seem to be supported by th e following characters of th e Ankylosauria : lack of contact between max illa with lacrimal ; presenc e of th e postfrontal an d cpipterygoid ; (?) retention of th e so called "tabular" in the posterior margin of skull roof; en ormously large external nasal ope nings - reaching far over th e skull ro of, seco nda ry covering of skull roof by derm al elem ent s ; loss of teeth in fr ont of th e maxilla an d dent ary ; and exte nsive armour. Critical remarks of THULBORN (1972) concerni ng th e similarities suggested by WALKER (1961) refer on ly to orn itho­ pod ornithisc hians and do not argus against an aetosaurian ances try for anky losaurs . Di versity of pelvic structure in Actosauria (e. g. th e small pubis in Aetosaurus erassieauda, stro ngly pro­ longed anterior process of th e ilium an d short posterior one in Typothorax meadei ) as well as th e variability of the position an d dimensions of temp oral fenestrae support the possibility of a connection with ankylosaurs. This d oes not mean that k nown aetosaur might be an a ncestor of the ankylosaur line, bu t an aeto sauria n origin for ankylosaurs is no less acceptable th an an orn ithopod origin. The concept of a monophyletic origin of orni thischians (e.g.T HULBORN, 1971, 1975), or extreme hyp otheses of monophyly of all th e Dinosauria (BAKKER & G ALTON, 1974) are not convincing. The stateme nt by THULBORN (1975, p. 251)"The lack of fossil evide nce makes it impossible to maintain th at anyone th eory of Ornithi schia ancestry is more accept abl e th an an other", is appropriate . His stateme nt (I.e.) say ing th at severa l lines of thecod onts during the Triassic peri od approached th e dinosau rian level of organization an d th at th e observed similarities betw een vari ou s dinosaurian groups are a result of parallel evolution within th ose lines, seems to be m ost convincing. The Scelidosauridae (Seelidosaurus and Sareolestes) might also be assigned to th e ankylo­ saurid line of Ornithisch ia (see fig. 13). Lack of detailed description of Scelidosaurus (specim ens BM(NH) R. 1111 and BM(NH) R. 6704, CHAR IG, 1972) prevents determination of the systematic position of th e Scelidosau ridae. Suggestion s that Scelidosaurus is a primitive represent at ive of th e An­ kylosauria (ROMER , 1968) is, at this moment, gr ou ndl ess, as lack of an anterior pubic process has been stated for all early Ornithisch ia. It follows that the lack of thi s process in Seelidosaurus do es not speak for its assignment to th e Ankylosauria. Reduction of digits in Scelidosaurus is mo re advanced than in k nown early Cretaceou s ankylosaurs, making such assignment even more difficult. Similarl y u nsatisfactory are arguments claiming assignment of Sc elidosaurus to th e Stegosauria. Assignment of Scelidosaurus to th e su border Ornithop oda, grade Brachypoda by THULBORN (1975) is also un acceptabl e. His opinion (1975) : " In practically every part of its skeletal anatomy S. harrisoni appears to be a perfectly acceptable orn itho pod din osaur" lack sufficient gr ounds. ANKYLOSAURIDAE FROM MONGOLIA 139

HABITS OF ANKYLOSAURIDAE

Keenly developed olfaction in ankylosaurids is probabl y connect ed with relative sluggishness of th ese animals. They were the slowest and clumsiest of all the Ornith ischi a. Not only the presence of armour, but also forelimb st ructure a nd confo rmation of th e pect oral girdle as well as development of limb mu scul ature indicat e lack of adaptatio ns to qu ick movem ents (COOMBS, MS). The position of articular surfaces on humerus, radius, and ulna indicates a strongly flexed attitude for th e forelimb of Saichania, with th e humeru s oriented outwa rd , stro ngly horizontally , an d slightly backward and the ulna and radius strongly obli que in a medi al direction. Fusion of th e corac oid with the scapula and an additional connecti on of the coracoid with the verte bral column indicat e lack of mobility in the pect oral girdle. A relatively extensive range of phalangeal movem ents both relat ive to th e metacarpals an d to one anothe r is cha­ racteristic of Saichania and othe r M on golian ankylosaurids . Extension movem ents prevailed over flexion. Because of the strongly flexed position of the fore limbs, the ante rior part of the trunk in Saichania was slightly raised off of th e groun d. The distance from the glenoi d cavity of Saichania chulsanensis (ZPAL MgD-I/l 12) to the grou nd is abo ut 35 em , The stro ng ventral armour of the bcdy indicat es a low position for the abdo me n. Differences in the structure of the humerus and the scapuloco ra coid in Saichania, Tarchia and Pinacosaurus indicat e that the humerus was most vertical in Pinacosaurus. The structure and curvature of th e pectoral girdle and ribs and the p osition of limbs suggest that the whole trunk of Saichania was flatten ed dorso-ventrally. In Pinacosaurus th e ante rior part of th e trunk was similarly flattened, wh ereas in Talarurus plicatospineus the trunk is slightly co mpressed laterally. The pelvic region was distinctly raised over the pectoral girdle in all specimens as in all anky losaurs, an d the hind limb was m or e vertical th an fore limb. The stru cture of th e occipi ta l cond yle indicates that th e atlas of Pinacosaurus an d Talarurus embraces the condyle ventro-posteriorl y wh ich gave good vertical and horizontal mobility to the head. Strong obl iqu e displacem ent of anterior articular surfaces of the cervical vertebrae in Pinacosaurus and posterior surfaces in Talarurus offer a consid erable ran ge of vertical movements to th e neck. In Saichania the articular surface of the occipita l condyle is develcped only ventrally.• Much smaller displacem ent of th e articular surface of verte brae than in Pi­ nacosaurus, together with a ventra l occipita l co ndyle, prove lesser vertical mobility of the head and neck in Saichania. The range of horizontal movem ents was much greater than th at of ver­ tical ones. Coossificati on of th e ribs of the first dor sal vertebra with the pectoral girdl e, the ch aracter of zygapophysal connections and coossification of distal dorsal vertebrae indicate little mobility in the back. The stru cture of th e caudal vertebrae in Pinacosaurus, Talarurus and " Dyoplosaurus" indi cat es that both lat eral and vertical mobility of th e tail was possible only in the proximal part between th e third to fourteenth caudals. In Pinacosaurus, because of very high neural spines in th at region , the vertical m ovem ents wer e also limited. Coossification of neural and hemal arches distall y in th e tail, and fusion between distal vertebral centra, indi cate that th e posterior porti on of th e tail for mi ng th e tail-club was immovable. In connection with limited general mobility, olfactio n probabl y play ed a great role in the life of ankylosaurs, especia lly in sea rching for food, find ing a mat e, and det ect ing an enemy. Presence of th e Jacobson's organ proves a necessity to expand the anterior olfac tory area of the snout, in associa tion with procurement of a particular food. A well developed hyoid ap­ paratus and a stro ngly developed entoglossal process may indicate the existe nce of a long, thin, and movabl e tongue, similar to som e recent Squamat a. It is a common belief that ankylosaurs were entirely herbivor ou s (see COOMBS, MS). It seems highly probabl e, however , that they may have taken animal food as well, such e. g. as insect s and their larvae (NOPCSA, 1928), which co uld have been dug out with the long tongue. Occ asion al feedi ng on carrion cannot be exclude d eithe r. Taking suc h a food is not denied 140 TERESA MARYANSKA by such features of the dentition as: small dimensions of teeth, weak development of enamel, and lack of possibilities of propalinar moveme nts of the mandibles. Th e prese nce of a long tongue and vomeronasal organ are arguments again st only herbivorous diet of ankylosaurids. MALEEV (1954) suggested that Syrmosaurus (= Pinacosaurus) may have dug itself in the sand by means of anteroposterior body movements. Th e body shape of Saichania, which is strongly dorso-ventrally flattened, armoured on the ventral side, and provided with sharp keeled plates on the lateral sides seems to support this hypothesis. The bony eyelid of Euoplocephalus described by COOMBS (1972) may have protected the eyes when the animal dug in the sand. The structure of the manus and pes, which are terminated with nail or hoof-like phalanges, contradicts this view to some extent. Th ese were not fossorial limbs. Hence some ankylosaurs may have dug by body movements as pr oposed by MALEEV (1954). Modification of the scapulocoracoid observed in Hyleosaurus by COOMBS (MS) suggests fossori al habits for the Nodosauridae and may also point to similar fossori al tendencies in all Ankylosauria. In my opinion the habit s and food of the ancestors of ankylosaurs were similar to those of the present-day Ornith orhynchus. Th ey should be looked for among the Triassic Reptilia .with a terrestrial-aquatic mode of life, and not among Triassic Ornitischia which were not only bipedal but also cursorial animals.

Polska Akademla Nauk Muzeum Z iemi 00-488 Warszawa, AI. Na Skarpie 20/16 July, 1976

REFERENCES

BAKKER , R . T . & G ALTON, P. M. 1974. D inosaur rnon op hyly an d a ne w class of verteb ra tes . - Nature, 248, 168-172. BARSBOLD, R. (EAPCEOJIA, P .) 1972. Bnocrpar nrparpna I I npcc ncnornn.re MOJIJIIOCI

India, 30, 75- 79. - 1935. Is Lametasaurus indicus an armoured di nosaur? - Amer. J. Sci,. 5, 30, 138-141. CHARIG, A. J . 1972. The evolution of the archosa ur pelvis and h ind-limb, an ex plan ation in functional te rms. In: "Studies in vertebrate evolution", (K. A. JOYSEY & T. S. K EMP ed s), 121- 155, Edinburgh. - 1973. J urassic and Cretaceous dinosaurs. In : "A tlas ofpa leobiogeography" , (A .H ALLAMed .), 339-352. A msterdam­ London-New York. - ATRRIDGE, J. & CROMPTON, A. 1964. On the origi n of the sauro pods and the classification of the Sa urischia . - Proc. Linn. Soc. London, 176, 197-221. CHENG, C. C, T ANG, Y. J., CHIU, C S. & YEH, H . K. 1973. N ot es on th e Upper Cretaceous -Lower Te rtiary of the Nanhs iun g Basin, NK wantung. - Vertebr. Palasiat. 11, I, 18- 28. COLBERT, E. 1945. T he hyoi d bon es in Protoceratops and in Psittacosaurus. - Amer. Mu s. Novit ., 1301, 1-10. CooMBS, 1972. W. P. 1972. The bony eyel id of Euoplacephalus (Re ptilia , Ornithischia). - J. Paleont, 46, 5, 637-650. ANKYLOSAURIDAE FROM MONGOLIA 141

-M. S. The Ankylosauria (Reptilia, Ornithischia). - Unpubl. Ph. D. thesis, Columbia University, Dept. of Biology, 1-306, New York. DASHZEVEG , D. 1975. New primitive therian from the Early Cretaceous of Mongolia. - Nature, 256, 5516, 402---4{)J. DONG Zm-MING, 1973. Dinosaurs from Wuerho. - Mem . Inst. Vert. Paleont. and Paleoant., 11, 45-52. EATON, TH. H. 1960. A new armored dinosaur from the Cretaceous of Kansas. - Univ. Kansas Paleont. Contrib. Vertebr., 8, 1-24. EFIMOV, M. B. ( E~ II Mo B , M. E.) 1975a . Xaancoaanpnn 113 HiD!bird from the Upper Cretaceous of Mongolia. Results Pol. Mong. Pal. Exp. V. - Palaeont, Pol., 30, 103-109. FORBRINGER, M. 1922. Das Zungenbein der Wirbeltiere insbesondere der Reptilien und Vogel. - Abhandl. Heidelberg Akad. Wiss., B, 11, 1-164. GALTON, P. M. 1973. Redescription of the skull and mandible of Parkosaurus from the Late Cretaceous with comments on the family Hypsilophodontidae (Ornithischia). - Life Sci. Contr., R. Onto Mus., 89, 1-21. - 1974. The ornithischian dinosaur Hypsilophodon from the Wealden of the Isle of Wight. - Bull. Brit. Mus . (Nat. Hist .), o-«, 25, I, 1-152. GILMORE, CH. W. 1914. Osteology of the armored dinosauria in the United States Nat. Mus., with special reference to the genus Stegosaurus. - Bull. U. S. Nat . Mus., 89, 1-143. - 1923. A new species of Corythosaurus with notes on other Belly River dinosaurs. - Can. Field Nat., 37, 46--52. - 1930. On dinosaurian reptiles from the Two Medicine Formation of Montana. - Proc. U. S. Nat. Mus., 77, 1-39. - 1933a. On the dinosaurian fauna of the Iren Dabasu Formation. - Bull. Amer. Mus. Nat . Hist., 67, 2-78. - 1933b. Two new dinosaurian reptiles from Mongolia with notes on some fragmentary specimens. - Amer. Mus. Novit., 679, 1-20. - 1943. Fossil lizards of Mongolia. - Bull. Amer. Mus. Nat . Hist., 81, 4, 361-384. GOODRICH, E. S. 1930. Studies on the structure and development of vertebrates. - MACMILLAN & Company, 1-837 London. GRADZINSKI, R. 1970. Sedimentation of dinosaur-bearing Upper Cretaceous deposits of the Nemegt Basin, Gobi Desert. Results Pol. Mong. Pal. Exp. II. - Palaeont. Pol., 21, 147-229. - KAZMIERCZAK, J. & LEFELD, J. 1969. Geographical and geological data from the Polish-Mongolian Palaeontological Expedition. Results Pol. Mong. Pal. Exp. I. - Ibidem, 19, 33-82. -& JERZYKIEWICZ, T. 1972. Additional geographical and geological data from the Polish-Mongolian Palaeontological Expeditions. Results Pol. Mong. Pal. Exp. IV. - Ibidem, 27, 17-32. HASS, G. 1969. On the jaw muscles of ankylosaurs. - Amer. Mus. Novit., 2399, 1-11. HEATON, M. J. 1972. The palatal structure of some Canadian Hadrosauridae (Reptilia: Ornithischia). - Can. J. Earth . Sci., 9, 2, 185-205. HlJENE, F. 1956. Paleontologie und Phylogenie der Niederen Tetrapodern. VEB Gustav Fischer Verlag. 1-716, Jena. KALANDADZE, N. N. & KVRZANOV, S. M. (KAJIAll):lA):l3E , H. H. & KVP3AHOB, C. M.) 1974. HmI

-& BARSBOLD , R. 1972. Narrat ive of the Polish-M ongolian Palaeon tological Expeditio ns 1967-1971. Result s Pol. Mong. Pal. Exp. IV.- Ibidem, 27, 5-13. KONZHUKOVA, E. D.(l{OI/)HYITyrannosauridae. - CODMecTIlaH COBeTcHo-MolIro.TJhCHaH fI aJICOH­ TOJIOrJl' teC!

- & BARSBOLD, R . 1972. A new d inosaur , Gallimimus bullatus n. gen. , n.sp. (Orn itho mi mi dae) from the Upper Cret ace­ ous of Mongol ia. Results Pol. Mon g. Pa l. Exp. IV. - Ibidem , 27, 103-143. OSTROM, J. H. 1961. Cranial morpho logy of the had rosau rian dinosaurs of North America . - Bull. A mer, M us. Na t. Hist ., 122, 2, 39-186. - 1962. The cranial cre sts 01 hadro saur ia n di nosaurs. - Postilla, 62, 1-29. - 1970. Stratigraphy and pa leonto logy 01 the Clover ly Formation (Lower Creta ceous) of the Bighorn Basin a rea, Wyoming and Montana. - Bull. Peabody Mus.Nat. Hist. 35, 1-152. PAR KS, W. A . 1924. Dyoplosaurus acutosquameus, a new gen us and species of armo red dinosaur ; with not es on a skeleton of Prosaurolophus maximu s. - Univ, Toronto Stud. Geo/. Se r., 13, 1- 35. PARSONS, TH. S. 1959 a. Studies on th e compara tive em bryo log y of the reptilian nose. - Bull. Mus. Compar. Zool., 120, 2, 103- 277. 1959 b. Nasal a na tomy a nd the ph ylogen y of re ptiles. - Evolutio n, 13, 2, 175-187. - 1967. Evo lution of th e nasal structure in th e lower tetrap oda . - Amer. Zoologist, 7, 397-413. - 1970. The no se and Jacobson 's organ. In : " Biology of the Rept ilia" , (c. GANS & TH. S. PARsoNS eds), B, 2, 99-186, Academic Press, London - New York. RIAIJI NIN, A. N. (PRElfHjlH, A. H. ) 1939. <:pa yHa n03BOIIO' llIbIX 113 nepxnero MeJIa toxcnor o Ka3aXCTaHa. - Tp. lI,HHTPH, 118, 1- 40. ROM ER,A. S. 1927. The pelvic mu scul ature of orni thisc hian dinosau rs. - Ac ta Zool., 8, 2-3, 225-275. 1956. Osteology of th e Reptilia . 1-772, C hicago. - 1966. Vert ebrate paleontology. - Un iv. Chicago Press, 1-468, Chicago and London. - 1968. NOles and co mme nts on Ver tebrate Pal eon tol ogy. - Ibidem, 5-304. ROZHDESTVENSKY, A. K. (PO)I{):(ECTBEECI{jlH, A . K .) 1955. H oni.re nannsre 0 ncnr r axoaanpax - MeJIOBbIX 0PHlITO­ nortax. - B onpocu eeon. A3UU, 2, 783-788. 1957. YTHOHOC bIH ):(1I1I03anp - 3aYP0,JICQJ 113 nepxnero MeJIa M OHrO)lIllI. - Vertebr. Palasiat .• 1. 2, 129-149. 196 1a . ITOJIe.B ble jICCJIe):(OBalljIR C JneTclParasaurolophus. - Pa/eont . Contr. R. ant. Mus., 3, 1-7. SAH NI, A. 1972. The vertebra te fauna of th e Judith Ri ver Formation , Montana. - Bull. Amer. Mus. Nat. Hist ., 147, 6, 325-412. SOCHAVA, A. V. (C OtIAnA , A . E.) 1975. Crparnrpadnra II ,JIIITOJIOrIIH nc pXHeMeJIOB bIX OTJIO}Hf IlIlH 10 )IAlberta . - Bull. Nat. Mus., Canada, 54, 28-35. 1942. New res toration of a hooded du ck-bill ed di nosaur. - Pa/eont. J., 16, 133-134. 1953. A new handro saur from th e Oldman formation of Albe rta , d iscussion 01 nomenclature. - Bull. Canada Dept . Resources Deve/op. , 128, 275-286. SUKHANOV, V. B. & NARMANDAKH, P.(C YXAHOB , B. B. & H APMAIIJ\AX, IT.) 1975. l.Iep enaxlI rpyrmsr Basilemys (Chelonia, Dermat emyd idae) B Asnn.-C OBMecTHaR ConeTcHo-MoHr OJIbCHaR ITaJIeOHTOJIOr UtIeCHali 3Hcne­ P;IIQlIlI, Tp. 2, 94-101 , Mocrcaa. SULIMSKI, A. 1972. Adamisaurus magnidenta tus n.gen., n.sp, (Sauria) from th e Uppe r Cretaceous of Mongolia. Results Pol. Mong. Pal. Exp. IV. - Palaeont, Pol. 27, 33-40. - 1975. Macrocep halosau ridae an d Polyglyphanod ontidae (Sa ur ia) from th e Late Cretaceou s of Mon goli a. Re sults Pol. Mong, Pal. Exp . VI. - Ibidem , 33, 25- 102. SWINTON, W. E. 1970. The dinosaurs. 1-331, London . SHUVALOV, V. F. (llIYn AJIOn, B. <:P .) 1974. 0 r eOJIOrlltIeCHOM crpoenmr II nospacre MeCToHaxQ}H):(eIlIIH X o6yp II X y­ P3H-,UYX. -C OBMecTllaR C OBeTcHo-MoHrOJIbCHaH ITa,JIeOnTOJIOr H'IeCHali 3HCne,llllQIUI, Tp. 1, 296-304, M o­ CHBa. 144 TERESA MARYANSKA

- 1975. Crparurparpa a Me3030H U eHTpaJIbHOH M OHr OJIlllI. - COBMeCTHaH COBeTCl< o-MoHroJIbCIIIbirds.- Austr. J. Zool. , 23, 249-270. TVERDOCHLEBOV, V. P. & TSYBI N, J. I. (T BEP,l:\O XJIEBO B, B. n . & U hIBIlH, 10. H.) 1974 . Teneanc BepXHeMeJIOBbIX Me CTOl1aXO>I<,l:\e HIlH ,l:\llH03aBpOB T yrpmcna-yc II AJIar T 3r . - COBMeCTHaH COBeTcHo-MoHrOJIbCHaH Ilaneoaro­ noru-recxaa 3 Ifossils collected by Yu an from Sinkiang and Inner Mongolia. - Vertebr. Pa/asiat., 8, 4, 398--401. - 1965. Note on the reptilian remains from Nanhsiun g, K wantung. - Ibidem, 9, 3, 292-297.

EXPLANATION OF PLATES

PLATE 19

Paso

Pinacosaurus grangeri GILMOR E, 1933 ...... 101 (see also pis 20-25)

Upper Cretaceous, Djadokhta Formati on, Bayn D zak, Gobi D esert, Mongolia ; ZPAL MgD-I1/1

1a. Skull of a young indiv idua l, lateral view. 1b. The same specimen, occipital view . 1c. The same speci men, palatal view.

All x ca 0.5

Photo : M. Malachowska-Kleiber ANKYLOSAURIDAE FROM MONGOLIA 145

PLATE 20

Pinacosaurus grangeri GILMORE, 1933 ...... 101 (see also pIs 19 and 21-25)

Upper Cretaceous, Djadokhta Formation, Bayn Dzak, Gobi Desert, Mongolia; ZPAL MgD-II{1

I a. Skull of a young individual, anterior view, x ca 0.5. I b. The same specimen, dorsal view, x ca 0.5. 2. Left premaxilla in lateral view, three nasal openings are visible, x ca I. 3. Ventral part of the endocranial cavity, dorsum sellae and pituitary fossa are visible, x ca I. 4. Right basicranial region in oblique latero-dorsal view, the right stapes and opisthotic not in natural position are visible, x ca 2. Photo: M . Malachowska-Klelber

PLATE 21

Page Pinacosaurus grangeri GILMORE, 1933 101 (see also pls 19-20 and 22-25)

Upper Cretaceous, Djadokhta Formation, Bayn Dzak, Gobi Desert , Mongolia; ZPAL MgD-II{1

1a. Right maxilla in lateral view, x ca 1.5. 1b. The same specimen, medial view of the maxillary teeth, x ca 1.5. 2a. Maxillary teeth, labial surface, x 4. 2b. The same specimen, lingual surface, x I. 3. Posterior part of palatal region, the palatines and contact of the vomer with the pterygoid are visible, x ca 2.

Photo: M. Malachowska-Klelber

PLATE 22

Page Pinacosaurus grangeri GILMORE, 1933 ...... 101 (see also pIs 19-21 and 23-25)

Upper Cretaceous, Djadokhta Formation, Bayn Dzak, Gobi Desert, Mongolia ; ZPAL MgD-I{1

I. and 5. Middle dorsal vertebrae. 2a. The neural arch of the middle dorsal vertebra, lateral view. 2b. The same specimen, posterior view. 2e. The same specimen, anterior view. 3. The isolated cervical vertebra, posterior view. 4. The neural spine of the posterior dorsal vertebra, lateral view. 5. Three dorsal vertebrae from the presacral rod ; lateral view.

10 - Palaeontologia P ol onica N o. 37 146 TERESA MARYANSKA

6a. Mandibular teeth, labial side, x 3. 6b. The same specimen, lingual side, x 3. 7a. Left mandible with symphyseal portion and predentary, lateral view. 7b. The same specimen, ventral view. All x 0.5, except fig. 6 Photo : M. Malachowska-Klelbe

PLATE 23

Page Pinacosaurus grangeri G ILMORE, 1933 101 (see also pIs 19-22 and 24-25)

Upper Cretaceous, Djadoktha Formation, Bayn Dzak, Gobi Desert, Mongolia; ZPAL MgD-II/1

1a. Right humerus, dorsal view, x 0.5. 1b. The same specimen, ventral view, x 0.3. 1c. The same specimen, posterior view, x 0.3. 2a. Proximal portion of the right ulna, postero-medial view, x ca 0.3. 3a. Proximal portion of the right radius, antero-medial view, x 0.3. 3b. The same specimen, posterior view, x ca 0.3. 4a. Left scapula, ventral view, x ca 0.5. 4b. The same specimen, dorsal view. 4c . The same specimen, internal view. 4d. The same specimen, external view. Sa. Right coracoid, internal view, x ca 0.5. 5b. The same specimen, externa l view. 6a. Latera l portion of the cervical derma l half-ring , external view, x ca 0.3. 6b. The same specimen, internal view. Photo: M. Mo lachowska-Kteibe

PLATE 24

Page Pinacosaurus grangeri GILMORE, 1933 ...... 101 (see also pIs 19-23 and 25)

Upper Cretaceous, Djadokhta Formation, Bayn Dzak, Gobi Desert, Mongolia; ZPAL MgD-II/1 1a. Left femur, posterior view, x ca 0.3. 1b. The same specimen, anterior view. 1c. The same specimen, internal view. 2a. Left tibia, posterior view, x ca 0.5. 2b. The same specimen, anterior view. 3a. Left fibula, lateral view, x ca 0.3. 3b. Th same specimen, medial view. 40. Two sacral vertebrae, dorsal view, x ca 0.3. 4b. The same specimen, ventra l view. Sa. Right metacarpalia I-IV and proximal articular surface of metacarpal V in articulation with preserved pha­ langes; anteri or view; ZPAL MgD-II/9. X ca 0.7. ANK YLOSAURIDAE FROM MONGOLIA 147

5b. The same specimen; posterior view. 6. Damaged left metatarsalia I-IV with preserved phalanges, anterior view; ZPAL MgD-II/9, x ca 0.5. 7. Terminal phalanx of the left metatarsal III, ventral view; ZPAL MgD-II/9, x ca 0.5. Photo: M . Malachowska-Kleiber and L. Dwornik

PLATE 2'

Page Pinacosaurus grangeri GILMORE, 1933 ...... 101 (see also pIs 19-24) Upper Cretaceous, Djadokhta Formation, Bayn Dzak, Gobi Desert , Mongolia la. Left ilium, dorso-Iateral view; ZPAL MgD-II/I, x ca 0.5. 1b. The same specimen, internal view. 1c. The same specimen, ventro-lateral view. 2a. Left ischium, external view; ZPAL MgD-II/I, x ca 0.3. 2b. The same specimen, internal view. 3. Left cervical rib, lateral view; ZPAL MgD-II/I, x ca 0.4. 4a. The chevron of the middle caudal, posterior view; ZPAL MgD·II/l, x ca 0.5. 4b. The same specimen, lateral view. 5. Anterior portion of the tail-club, dorsal view; ZPAL MgD-II/9, x ca 0.3. Photo : M . Mala chowska -Kleiber

PLATE 26

Pace Talarurus disparoserratus (MALEEV), 1952 ...... 100

Upper Cretaceous, Bayn Shireh svita, Sheeregeen Gashoon, Gobi Desert, Mongolia la. Left maxilla with preserved teeth, internal view; Holotype, PIN 554/1-1, x ca 0.8. 1b. The same specimen, lateral view.

Page Talarurus plicatospineus MALEEV, 1952. 100

Upper Cretaceous, Bayn Shireh svita, Bayn Shireh, Gobi Desert , Mongolia 2. Left ulna, antero-lateral view; PIN 557-80/1, x ca 0.3.

Page Pinacosaurus grangeri GILMORE, 1933 ...... 101 (see also pIs 19-25) Upper Cretaceous, Djadokhta Formation, Bayn Dzak, Gobi Desert, Mongolia 4. Left humerus, dorsal view; PIN 614-1/24, x ca 0.3. 3. Left ulna, antero-dorsal view; PIN 614-1/10, x ca 0.3. Sa. Right scapulocoracoid, internal view; PIN 614-1/7, x ca 0.3. 5b. The same specimen, dorsal view. Courtesy 0/ PIN, photo W. Sknriyriskl 10· Palacontologia Polonica, No. 37, / 977 Pl. 19

T. MARYANSKA: ANKYLOSAURIDAE FROM MONGOLIA Palaeontologia Polonica, No. 37, / 977 PI. 20

T. MARYANSKA : A NKYlOSAURIDAE FKOM M ONGOLIA Palaeontologia Polonica, No . 37, 1977 PI. 21

T. MARYANsKA : ANKYLOSAURIDAE FROM MONGOLIA Palaeontologia Polonica, No. 37, / 977 Pl. 22

T. MARYANsKA : ANKYLOSAURIDAB FROM MONGOLIA Palaeontologia Polonica, No. 37, / 977 PI. 23

T. MARYA:'ISKA : ANKYLOSAURIDAE FROM MONGOLIA PI. 24 Palaeontologia Polonica, No. 37, 1977

T. MARYANSKA : ANKYLOSAURIDAE FROM MONGOLIA Palaeontologia Polonica , No . 37, / 977 PI. 25

T. MARYANsKA . AN KYLOSAU RIDAE FRO M MONGOLIA Palaeontologia Polonica, No. 37, 1977 Pl. 26

4

T. MARYANSKA: ANKYLOSAURIDAE FROM M ON GOLIA Palaeontologia Polonica, No. 37, 1977 PI. 27

T. MARYANSKA : ANKYLOSAURIDAE FROM MONGOLIA Palaeontologia Polonica, No . 37, / 977 Pl. 28

T. MARYANSKA: ANKYLOSAURIDAE FROM MONGOLIA Palaeontologia Polonica, No. 37, / 977 PI. 29

T. MARYANSKA: ANKYLOSAURIDAE FROM MONGOLIA Palaeontologia Polonica, No. 37, 1977 PI. 30

T. MARYANsKA: ANKYLOSAURIDAE FROM MONGOLIA Palaeontologia Polonica, No. 37, 1977 PI. 31

T. M ARYANSKA : ANKYLOSAURIDAE FROM MONGOLIA Palaeontologia Polonica,No. 37, 1977 PI. 32

T. MARYANSKA: ANKYLOSAURIDAE FROM MONGOLIA Palaeontologia Polonica,No. 37, 1977 Pl. 33

T. MARYANSKA : ANK YLOSAURIDAE FROM MONGOLIA Palaeontologia Polonica, No. 37, 1977 Pl. 34

T. MARYANSKA: ANKYLOSAURIDAE FROM MONGOLIA Palaeontologia Polonica, No. 37, 1977 Pl. 35

T. MARYANSKA : ANKYL.OSAURlDAE FROM MONGOLIA Palaeontologia Polonica, No. 37, 1977 PI. 36

T. MARYANSKA: ANKYLOSAURIDAE FROM MONGOLIA